Cream of the Crop 1/Cream of the Crop 1.iso/PROGRAM/TP032292.ARJ/03-22-92.TPC

Text File | 1992-03-22 | 151KB | 4,300 lines

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 06:53:26
From       Trevor Carlsen
To         Derek Westcott
Subject    Tp 6.0 Ide



 DW> I have to disagree with you on that one.  The IDE provides at least
 DW> one necessary debugging feature - location of runtime errors in the
 DW> code...

That is available in any good programming editor - not just the IDE.

 DW> Outside in DOS, all you get is a meaningless address...when you
 DW> load up something to find that address, you've got to rerun the code

 DW> under the finder utility...

You would appear not to have used Brief, VEdit, Multi-Edit, Unity, QEdit etc.
etc. when they are properly set up.

TeeCee



--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/15  15:45:21

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 07:58:00
From       Trevor Carlsen
To         Jud Mccranie
Subject    Re: New Reader!



 JM> I'm pround of the fact that I've never used assembler,
 JM> and I probably never will...

Fine... however there are occasions when such an attitude is going to result
in bloated and/or slower code than is otherwise necessary.  To my way of thinkin
 a serious programmer has to make use of every possible tool or means at his/her
disposal to produce product that is professional and slick.  Restricting yoursel
 in the way you state will prevent that final polish.

 JM> I wish that was true foe me too, however I know for
 JM> sure at least 2 are currently using 8088.

I now restrict myself to coding *only* for '286 or better machines.  Now before
all those out there scream and say how silly that is, I am in a situation
totally different to most and therefore such a policy works for me.  For most
people, and obviously this applies to you, it wouldn't be possible.  However
nothing stops you producing two versions.

 JM> Because a major program needs the IDE MORE than a Q&D.  Fast
 JM> compilation/editing/debugging.  The integrated debugger is a
 JM> fantastic tool that you don't have anything remotely
 JM> comperable (that I know ov) anywhere else.

I agree... it is a fantastic tool.  However the point I make is that it is
not really necessary to forego this.  TD is more powerful and, given the right
setup with a programmer's editor, almost as easy to use.

How do you use the integrated debugger so frequently that "almost" becomes
a problem?  My work style involves debugging small units thoroughly and then
any major problem is in new work. As you can't use assembler, the ID's lack
of a disassembler would not be as important to you as it is to me.  The disassem
led code is a great way of seeing faults in your high-level code.

Another couple of points I would raise is your design method. It appears from
your comments that your big work is developed in an ad hoc way.  If I am wrong
please don't be offended by this comment, the reason I say this is because
of comments I see from you about several updates per week, constant use of
the debugger, not being aware of the uses and value of exit procedures etc.

What is your big application used for and how many pages of documentation
(particularly the user manual) are involved?

In any big project the design stage is by far the hardest, longest and most
important.  Normally most houses would budget on this stage taking  60-80%
of the total project time.  Coding and obvious debugging probably represents
less than 20% of the time and effort with the remainder being in-house testing
and acceptance testing and ironing out any bugs this uncovers.  By design
stage I mean analysis, functional specifications, general design (screen layouts
and design, program modules, flow etc.), user manu
als and test specifications. Only after this is done, and done properly, is
any coding done.  Obviously, as coding proceeds there may be some minor cosmetic
changes needed but any major hitch at this stage requires a proper procedure
for handling this to be implemented.

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/15  15:45:21

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 07:40:24
From       Trevor Carlsen
To         Jud Mccranie
Subject    Re: Tp 6.0 Ide



 JM> I disagree on the last part.  If I test in the IDE with range check
 JM> on, then the only thing that can cause it to run differently as an
 JM> EXE is an unitialized variable.  And that could make it run
 JM> differently on different runs whether it was in the IDE or an EXE, as
 JM> far as I know.

In any testing a fundamental rule involves testing the limits, upper and lower,
of all data.  This involves stack limits, heap limits, the whole works.  In
the IDE it would be difficult, if not impossible to do this properly.  TSRs
and ISRs cannot really be tested at all from the IDE unless you are fond of
rebooting!

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/15  15:45:21

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 09:12:51
From       Trevor Carlsen
To         Jud Mccranie
Subject    Re: New Reader!



 TC> Nothing stops you producing two versions.

 JM> I send out 1 or 2 new versions a week.  Not worth it for the cunfusion
 JM> for such a small benefit.

Are you serious?

 TC> I don't think I am aware of ONE thing that they cannot do that
 TC> the IDE does (apart from display a Borland Copyright notice :-)
 TC> ), and they have MANY functions not available in the IDE - some
 TC> insignificant but some really major.

 JM> Do they have the integrated debugger?  Case closed.

No they don't... but the case is not closed.  Good design should minimise
the use of the debugger to only the relatively harder bugs.  Then TD comes
into its own.  And it is still only one keypress away.


TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/15  15:45:22

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 09:18:13
From       Trevor Carlsen
To         Jud Mccranie
Subject    Re: Longer files in TP5



 TC> The TP5 IDE is severely crippled.  The TP6 IDE will allow you
 TC> to edit large files and its help system is significantly larger
 TC> and better than that in TP5. (Mind explaining why you consider
 TC> the help system to be inferior?)

 JM> He shouldn't be using single files that large anyway.  ...

Why?  I agree that the source code should never approach that size in a normal
situation but it is normal to expect that comments and documentation in the
source file to exceed the actual source size.  Also there are *many* times
you will want to load and read/edit much larger files that may not actually
be part of the program.

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/15  15:45:22

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 09:22:43
From       Trevor Carlsen
To         Jud Mccranie
Subject    Re: Pascal, The Language



 TC> Power and flexibility are usually mutually exclusive and
 TC> therefore should not be grouped together like that.  Assembler
 TC> is *by far* the most powerful language one can use.  NOTHING
 TC> comes a close second.

 JM> You are equating Power to speed of execution and size of the
 JM> EXE, I beleive.  ...

Not completely.  I define power as the ability of a language to do something.
 There is nothing that can be done in a high level langauge that cannot be
done with assembler, albeit it with more difficulty.  The reverse does not
apply, therefore assembler is the more powerful.

 JM> ... Try expressing a complex algorithm in assembler
 JM> and see how powerful it is.

I didn't say "it is easier to do in assembler", I said it is by far the most
powerful.  High level languages are primarily there to help the programmer
avoid the drudgery and difficulties of using assembler.  The power of a language
is not how easy it is for a programmer to achieve something in that language
but how efficient and capable the language is in executing that something.


TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/15  15:45:22

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 09:31:16
From       Trevor Carlsen
To         Kevin Higgins
Subject    Run Time Error 162



 KH> Is there anything thing I could do in my code which would
 KH> cause a run time error 162 (hardware failure)? I got a
 KH> report of this from a beta tester on a program after I
 KH> implemented the sort routine you helped me with (it works
 KH> fine on most machines -- this may be an isolated instace,
 KH> but the guy's system is sound...) Using the program with my
 KH> old, inefficient sort, it works fine, but with the new quick
 KH> sort you showed me he gets this error.... anything come to
 KH> mind?

Without being able to get my hands on the program I could not tell.  I would
suspect some kind of memory overwrite.  By memory the routine I supplied you
with could do this, even with range checking enabled, so you need to be absolute
y sure of preventing out-of-range errors.

All the usual caveats apply, watch unitialised pointers, single step through
in the debugger etc. etc.

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/15  15:45:23

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 09:37:57
From       Trevor Carlsen
To         Thomas Groff
Subject    Mucking about with disk files.



 TG> I recently saw a message someone posted asking how to slip some
 TG> data into the start of a file that already contained data and
 TG> it got me to thinking about disk files in general.
 TG> Wouldn't it be possible to: 1. create a file containing new data.
 TG> 2. delete an old file (The one you want to append.)
 TG> 3. twiddle the disk's file allocation table and directory entry to
 TG>    use disk space occupied by the erased file for the new file.
 TG> 4. reopen the new file and push the file pointer to include the
 TG>    erased files data.

I'd forget that idea very quickly if I were you!  It would only work if the
data to be added exactly matched the size of a cluster and even then would
be tricky and dangerous.  Not worth the effort.

 TG> Has anyone tried this?  Or does DOS not allow one to specify
 TG> Just what disk area is going to be used next.

No and no.

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/15  15:45:23

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 09:42:31
From       Trevor Carlsen
To         Robert Kantor
Subject    Help!



 RK> I am trying to utilize a two dimensional array.  The first field will

 RK> contain numbers the second a string of 12 characters.  Does anyone
 RK> have a routine which will take the first field (numbers) and sort it

 RK> and the corresponding strings will be sorted too?

Better that you use a singly dimmensioned array of records to achieve that.

type
  YourRecord = record
                 number : word;
                 TheStr : string[12];
               end;

var
  YourArray : array[min..max] of YourRecord;

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/15  15:45:23

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 09:48:19
From       Trevor Carlsen
To         John Reid
Subject    Archiver



 JR>  * SMRead 3.0  #0286 *
      ^                   ^

John you may not be aware of it but your reader/editor is placing hi-bit charact
rs in the indicated positions.  Please fix if possible.

Trevor Carlsen
Moderator.



--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/15  15:45:23

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 09:53:01
From       Trevor Carlsen
To         Andrew Broughton
Subject    Variable-Length Records



 AB> Does anyone out there know of a (fast) way to get to a specific record

 AB> in a file that has variable-length records? What I have is a file with

 AB> sorted records, each ending in the same terminator, but of varying
 AB> length. What I want to do is be able to do a binary search on the
 AB> file, but I need a way to
 AB> jump to specific records.

It is simple to do by creating an index file.  Do this by reading through
the file and recording in the index file the location of each terminator.
 If this is still not sufficient ask again.

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/15  15:45:24

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 10:01:22
From       Trevor Carlsen
To         Jud Mccranie
Subject    Re: New Reader!



 RM> In some cases I have been required to use a particular editor
 RM> to conform with team requirements, specifically with version
 RM> control or standard metrics.  This is a contract stipulation
 RM> that is common for commercial programmers.

 JM> Not to detailed specs.  They tell me what they want it to do and
 JM> I do it.  Then if they want it changed I change it.  I don't have
 JM> to conform to team requirements.

A 400k .exe is a big program and could easily represent man-years of effort
by a programmer.  Such a program would be highly unlikely to be developed
by a one-man show unless the major part of that 400k is taken up by third
party library stuff.  As such and in total it probably still represents man
years of effort.

I would not consider contracting a big job with someone who cannot either
provide detailed written specifications or alternatively sit down with me,
or a team, and come up with those detailed specifications.  It is simply not
viable to quote a price without that being done.  And if I could find someone
stupid enough to contract me to do a big project on an hourly basis, I would
be in seventh heaven.  Successful businesses do not do that sort of thing!

 JM> Well, I can bet that if the program will fit in the TP IDE, I can
 JM> develop faster and better than anyone using Brief, QEdit, etc because
 JM> of the power of the integrated debugger.

I can place only one interpretation on that statement and that is that you
develop "as you go".  On smaller or unimportant jobs I guess that many of
us work that way and that would definitely make the ID a more important factor.
(I know that too often I work that way thinking that I can get away with the
slackness because of the nature of the task.)  On serious jobs though, that
method will produce poor code and poorer programs, necessitating frequent
revisions due to simple bugs or unplanned improvements.  I would be willing
to bet many dollars on a professsional programmer with a professional editor
and professional set of tools being more productive than if he was using the
IDE and the same tools.

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/15  15:45:24

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 13:31:17
From       Trevor Carlsen
To         Jud Mccranie
Subject    Re: New Reader!



 DM> I'm sort of curious why you'd be proud never to have used a
 DM> useful tool, but it's hardly a Pascal topic, so I won't ask.

 JM> Because I use only high-level languages.  I write app progs, not
 JM> utilites or system progs.  As far as speed, it is often easier to
 JM> write a fast one in a high-level language than assembler.  For
 JM> instance, an average version of quicksort in Pascal will easily
 JM> outrun a tight assembler implementation of bubblesort.

Of course it is and of course it will!  But you are comparing a Ferrari towing
a trailer to a small car.  Take the trailer off the Ferrari by using assembler
AND a QuickSort and there is once again no contest.  Sorting and searching
are excellent examples of code in ordinary everyday applications that benefit
substantially by the use of assembler, substantially enough to make a significan
 impact on the perception of the application by the end user.

TeeCee


--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/15  15:45:24

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 13:45:27
From       Trevor Carlsen
To         Jud Mccranie
Subject    Re: Tp 6.0 Ide



 JM> I've never seen a case where the program rn differently in the
 JM> IDE instead of EXE unless there was a dumb bug.

I'm not sure of this but do programs running in the IDE using the PrefixSeg
variable operate correctly?  (I have no way of checking without re-installing
the IDE)

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/15  15:45:25

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 13:48:44
From       Trevor Carlsen
To         Jim McNamee
Subject    TP 6.0



 JM> One inclusion I've been waiting for is dynamic or conformant arrays.


It would be a nice addition even though it is easy enough to pseudo implement
yourself using untyped parameters.

 JM> At the moment, to pass array data to a Proc or Function one must
 JM> explicitely use pointers or sprinkle typed constants throughout .TPUs

 JM> to assure that passed arguments are identical between called and
 JM> calling code.

I feel that this is not quite correct if my above suggestion is used.

 JM> ... The ISO standard would allow determination of array
 JM> size at run-time allowing arrays to be passed more easily.

Undoubtedly.

TeeCee



--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/15  15:45:25

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 13:54:45
From       Trevor Carlsen
To         Ken Koch
Subject    IDE's



 KK>  I just downloaded an IDE called UNITY310. This program looks
 KK>  great except for 1 thing. It does not let you specify the
 KK>  multiple directories, like the default TP 5.5 IDE does. Like
 KK>  setting up your include directories, or the object
 KK>  directories, etc. Is there any other IDE out there that
 KK>  provides this function.

Welcome to the club :-)  Your productivity will increase remarkably!  The
problem you are having is not due to Unity but to your config file for TPC.
 Set that up correctly and Voila!

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/15  15:45:25

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 13:58:55
From       Trevor Carlsen
To         Jud Mccranie
Subject    Re: Pascal Vs. C



 DN> Sounds good. Is Modula 2 more or less low-level capable than
 DN> TP?

 JM> Modula-2 is more low-level capable than Pascal...

True.  However if you were to say "Turbo Pascal" then I believe that the stateme
t is incorrect.

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/15  15:45:25

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 14:10:13
From       Trevor Carlsen
To         Mark Sandfox @ 970/201
Subject    Memory



 MS>   I have a large source code 400k+ that I can not compile.  I keep
 MS> recieving segment too large...

What is the actual error message?  Read the manual for information on the
actual error message you are receiving.

TeeCee



--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/15  15:45:25

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 14:13:47
From       Trevor Carlsen
To         Mark Sandfox @ 970/201
Subject    Files



 MS> How can I make a data file unreadable to the user?  Besides
 MS> incription.

Probably encryption is your only means available.

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/15  15:45:26

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 16:06:20
From       Trevor Carlsen
To         James Digiacomo
Subject    Dynamic Arrays in Turbo Pascal



 JD> the problem occurs when I try to display the contents stored in this

 JD> type of array. it seem to contain the data PLUS additional garbage
 JD> characters. As a reminder, it only happens when I compile my program

 JD> here's the declaration part of my program:

 JD> TYPE
 JD>      item = string;
 JD>      limittype = array [1..1] of item;
 JD>      numbertype= array [1..1] of item;

 JD> VAR
 JD>      Lim : ^limittype;
 JD>      temp, numbers : ^numbertype;

 JD> I use TP's GETMEM procedure to allocate the memory for these arrays
 JD> and after the program is finished, I release the allocated memory
 JD> using the procedure
 JD> FREEMEM.

You have not indicated the actual allocation and deallocation. That is the
important part.  Also show some examples of reading and assigning array element
values.

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/15  15:45:26

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 16:09:17
From       Trevor Carlsen
To         Bill Pantoja
Subject    Re: Files



 BP> If you wirte a fast encryption scheme (Other then a simple
 BP> NOT or XOR encryption), you can encrypt a 200MB file in
 BP> under 30 seconds.

I simply  refuse to believe that this is possible.  That represents 6.67 megabyt
s per second.  On a 486 33mHz even a simple no content loop does not execute
that fast.  eg
  for x := 1 to 200000000 do;
will take longer than 30 seconds.  On top of that you have all the necessary
[slow] disk accesses.

 BP> program I wrote for the Medical Board Of California, all the
 BP> data file were required to have a bullet-proof encryption
 BP> scheme.  Using a combination of NOT, random translation
 BP> tables, etc. I came up with a very good encryption scheme
 BP> which uses a negligable amount of time.  The data file is
 BP> over 2.7MB but by encrypting or decrypting the data in
 BP> memory rather then on disk, it speeds up the program and
 BP> protects against leaving the data "bare" on your disk if a
 BP> user reboots the machine.

About the only "bullet proof" encryption is what is known as a "one time pad"
method.  The encryption standard used by the US department of defense has
so far defied attempts to break it but no mathematical proof exists of it
being impossible.  It is also very slow on a PC.

Encryption of sensitive data can be made *almost* impossible to break without
recourse to heavy duty computer usage/power and plenty of time, skill and
imagination.  Non military or diplomatic data is rarely sufficiently valuable
to make the effort worth the expense and even then then time constraints may
defeat it.

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/15  15:45:26

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 20:11:08
From       Trevor Carlsen
To         Greg Williams
Subject    Generating sequential file names...



 GW> Does anyone have an efficient method of generating the next filename

 GW> in a sequence of names like the following:

 GW>         File000.ext
 GW>         File001.ext
 GW>         File002.ext
 GW>         .
 GW>         .

 GW> Where the number can be anything up to the whole eight characters
 GW> allowed for a filename (ie.  something like "00000231.ext")...
 GW> Up to now, I've been asking DOS to pass me a "unique" filename, but
 GW> that is no good if you want to see at a glance which is the latest...

Keep the next number + 100000000 in the sequence as a longint in a file and
just use TP's str procedure to produce the file name.

var
  fname : string;
  fnumber : longint;

  str(fname,fnumber);
  fname := copy(fname,2,8) + '.' + ext;

TeeCee



--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/15  15:45:27

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-16-92 06:27:03
From       Trevor Carlsen
To         Scott Hutchinson
Subject    Hiya



 SH>  randomize;
 SH>  writeln(random(1000));

 SH> This will generate a random number between 1 and 1000.  If you want it

 SH> to generate a number on a larger scale, use a number larger than 1000.

 SH>  If you have any more questions, I'll be glad to answer them for you.

Thanks...but try to answer them accurately! :-)

The above will display a random number ranging from 0 to 999.

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/16  19:51:36

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-16-92 16:22:49
From       Trevor Carlsen
To         Jud Mccranie
Subject    Re: Pascal Structure Philosophy



 JM> In the example he gave, he doesn't have to set the variable, and that
 JM> is one of his main points.  The variable is not checked until the end.
 JM> The "error" variable is set automatically when the file is opened.

The only time it makes any difference is at the start of the looping construct
and therefore it is of little or no consequence.  The easiest way is to start
the loop by an if statement and then place both checks at the end of a repeat
loop.  That way he avoids the required setting of the variable that he dislikes,
code size is identical and execution time is probably the same.

To set up a completely new and unnecessary looping construct is a strange
attitude.

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/16  19:51:36

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-16-92 17:33:47
From       Trevor Carlsen
To         Jud Mccranie
Subject    TP 6 Bad design



 JM> First is execution speed.  I believe I have the same options in
 JM> both, as far as checks, alignment, etc, however, I got the following
 JM> results on the exact same source:

 JM> TP 5.5    22.24 seconds
 JM> TP 6   G- 22.90
 JM> TP 6   G+ 22.91

 JM> 286 instructions made no difference, and 5.5 executes 3% faster than
 JM> either of of them.  Maybe I've done something different, but I don't
 JM> think so.  (I'm using Eagle's system on both.)

I decided to set up the TP6 IDE again so I could test this.  Obviously I cannot
comment on the actual results of yours as I don't know what your code was.


As you have stated elsewhere that all your testing is done in the IDE, the
above would be an invalid test if that is the method you used in your testing.


The TP6 IDE is an event driven environment and unless it releases ALL hooks
for that environment when executing a program from within itself, which I
do not think it does, a certain amount of time is needed to service those
hooks.  Because of this it is obvious, or should be to an experienced programmer
 that a program running from within the TP6 IDE will always be slower than
when the same code is executed from the command line.  That is exactly how
it turned out in my testing.

In the test my test program was 9% slower running in the TP6 IDE than when
executing from the DOS command line.  It was about 7.4% slower in the TP6
IDE than when compiled with TP5.5 and run from the command line.  However
when both versions were executed from the command line the TP6 version was
about 1.7% faster than the TP5.5 version.

I was unable to do a test running from within the TP5.5 IDE but I would have
thought a 3% figure might have been a little on the low side.  I think that
the TP5 IDE hooks some vectors also, perhaps the discrepancy may be there,
although without me running exactly the same set up as you it is a little
pointless to compare my results to yours. Suffice it to say that bench mark
testing from within different IDEs is invalid and I am very surprised that
someone with your stated experience and knowledge would do that (if in fact
you did, as appears likely from your results).

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/16  19:51:36

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-16-92 15:40:52
From       Trevor Carlsen
To         Gene Buckle
Subject    Re: Bbs Source.. / Range Checking



 GB> ... I am getting a range check error on one of his functions
 GB> that returns a boolean value.  I dies at "GetLogin := rcs"
 GB> where rcs is boolean (and is set to true or false depending
 GB> on what info is gathered in the body of the routine). It
 GB> works just fine if I $R-, but with range checking on, it
 GB> dies.

The easy way is to place the following lines immediately prior to the line
in question -

  if (byte(rcs) <> 0) and (byte(rcs) <> 1) then begin
    writeln('rcs has not been correctly initialised');
    halt;
  end;

I believe that you will find that rcs is not being initialised where you think
it is - perhaps an if-then-else-if-then statement's logic is off or a value
parameter is used where a reference parameter is needed.

Place a watch on it in the debugger to find out just when and how it is being
initialised.  The result may surprise you.

TeeCee


--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/16  19:51:37

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-16-92 15:49:34
From       Trevor Carlsen
To         Gene Buckle
Subject    Re: Bbs Source..



 GB> ... rcs is being initialized to FALSE at the beginning of the func.

Why not post the function?

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/16  19:51:37

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 16:21:00
From       Werner Berghofer
To         Jud McCranie
Subject    Longer files in Turbo Pascal 5




       Jud,

 > I con't consider TP 5 IDE "severely crippled" because of that.
 > In fact, it is BETTER that the editor limit the files to 64K.

       no doubt about the second statement, you are right.  But how else than
"severly crippled" would you call an editor which offers you Ctrl-Alt-Del
as only option if during editing (block copying etc.) the size of a source
file exceeds 64 KB?

       I'm very glad that Borland removed this not really seasonable editor
limit in version 6.0 of Turbo Pascal.

       W.

---
 * Origin: God is real... unless declared integer. (2:310/90.100)
 * Tossed by SFToss v1.00b on 92/03/17  04:34:44

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 08:22:31
From       Dj Murdoch
To         Jud Mccranie
Subject    Re: TP 6.0 IDE memory

  MO> BTW if you can't adapt to the 6.0 IDE, don't despair,
 MO> Borland Pascal++ is just around the corner ;-)

 JM> That's what I've heard.  I hope they take the users into
 JM> consideration this time.  They were headed in the wrong direction
 JM> w/ 6.0 IDE, IMHO.

It's probably too late to make any difference now, but what sort of changes
were you looking for?

--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/17  11:00:28

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 08:25:12
From       Dj Murdoch
To         Dale Frakes
Subject    Re: Binary Tree Linked Lists

  > Actually B-Trees are more effecient than Binary tress and
 DF>
 DF> I thought the B in B-tree stood for Binary.  What is the
 DF> difference between a b-tree and a binary tree?  (Just a
 DF> basic explanation, I don't want source and such).

B is for Balanced - to the extent possible, every node has the same number
of subnodes.  A binary tree can be so imbalanced as to be nearly a linked
list, if new nodes are always added on the right, for instance.

It's quite hard & time consuming to do adds & deletes to a balanced tree,
but the searches are fast.  A compromise method that allows easy adds and
deletes, but is 20 or 30 percent slower on searches, is a Skip List.  It's
a list with multiple indices:  there's one level that lists every node, the
next  level skips a few each step, the next one skips a few of those, etc.
 You start at the top level (fewest nodes) and do a linear search of 2 or
3 nodes to find the starting place at the next level.

The key thing that makes adds & deletes quick is that each new addition is
done completely independently of whatever is there.  You *randomly* choose
(according to a geometric distribution) how many levels a new addition should
fit into, and just add it to those.

Look in the June 1990 issue of the Communications of the ACM for an article
discussing these, or look on PDN Pascal nodes for SKIPLIST.ZIP containing
some sample code.

--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/17  11:00:28

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 08:37:04
From       Dj Murdoch
To         Ruurd Pels
Subject    Re: Turbo Vision Questions...

  RP> No sir it ain't! You are having a misconception about the
 RP> nature of TV. It is not a collection of things. TV is a
 RP> closely coupled class hierarchy containing a whole
 RP> template program, ready to run. Apart from the TCollection
 RP> class and it's descendants, it is ONE class hierarchy.

And apart from the TStream class and its descendants, and the TStringList,
and the TResourceFile.  All of which are useful in programs that have nothing
descended from TView.


--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/17  11:00:28

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 08:40:34
From       Dj Murdoch
To         Joshua Kersey @ 930/22
Subject    Re: Turbo Vision Questions...

  JK> Yes...but the fact remains that it still *IS* OOP, which
 JK> I'm not good with and can hardly apply with a good spray
 JK> bottle and a can of paste. I'd like to see Turbo Vision
 JK> WITHOUT the OOP, something I could actually use without
 JK> re-learning Pascal.

I can't imagine what TV without OOP would be.  Do you just mean a windowing
library?  There are lots of those.  The best would be Turbo Professional,
I've heard Turbo Technojock's Toolkit is good, and Blaise sells something.

--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/17  11:00:28

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 08:43:46
From       Dj Murdoch
To         Jud Mccranie
Subject    Re: Why I Like Tp 5.5 Ide (co

  JM> So the "O" should be BLACK - not cyan.  I look at "OK" and I think
 JM> the the hot letter should stand out.  In the other cases, the yellow
 JM> stands out from the black.  But in the case of "OK", with only two
 JM> letters (one cyan, one yellow), there is no way to tell which is
 JM> the hot letter unless you compare it to the rest of the screen, or
 JM> you rember "in that case, the yellow is the hot letter, and the
 JM> cyan one isn't" and the user shouldn't have to remember that.  He
 JM> should be able to tell at a glance which is the hot letter w/o having
 JM> to compate it with the rest of the screen or rember an arbitrary rule.

The trouble is that OK is special in two ways.  That's why it looks different
than any other button.  Seems smart to me...



--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/17  11:00:28

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 08:46:12
From       Dj Murdoch
To         Jud Mccranie
Subject    Re: TP 6 Bad design

  JM> I've been trying TP 6 again, and I've encountered even more
 JM> problems.

 JM> First, let me say that memory is no longer the hangup.  However,
 JM> everything else I've spoke of is still a problem, and I've
 JM> encountered several more pretty bad problems.

 JM> First is execution speed.  I believe I have the same options in
 JM> both, as far as checks, alignment, etc, however, I got the following
 JM> results on the exact same source:

 JM> TP 5.5    22.24 seconds
 JM> TP 6   G- 22.90
 JM> TP 6   G+ 22.91

$G+ can be slower than $G-, because of the bad implementation of ENTER and
LEAVE on the 80x86 family.  Borland should never have touched them.  That's
the only difference I would have expected.  Can you identify where the time
is being spent?

 JM> After you go through all of the tough song and dance to change an
 JM> option, if you don't OK it, the change doesn't take effect.  If
 JM> you F10 or escape from the menu w/o "OK"ing it, the change is lost.
 JM> This is an extremely bad design flaw.  The last thing you saw on
 JM> the screen should be the options in effect.

What do you mean "if you F10"?  That doesn't do anything in the menus on my
system.  The Escape behaviour is good, in my opinion.  After you've made a
pile of changes, it's nice to be able to accept them (Enter) or abandon them
(Esc).  Do you think it should ask if you really mean to abandon the changes?

 JM> When you escape from a option menu, it puts you back in the editor
 JM> rather than the menu.  This is bad design.

That's because options shouldn't be in menus, they should be in dialog boxes.
 Yes, the menu system in TP 6.0 isn't well designed.

 JM> With TP 5.x, I can hit F9 anywhere in the IDE to compile.  In 6.0,
 JM> you have to get out of a menu first.  This is very bad design.

There's a difference of philosophy here.  In a dialog box, the changes you
are making aren't applied until you hit Enter or Okay.  If you were allowed
to do a compile while in this state, you'd be compiling with the old options.
 Perhaps there should be a "Accept all options and compile" hot key, but people
mostly don't spend a lot of time flipping options, so it's probably not worth
adding.

--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/17  11:00:28

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 09:00:29
From       Dj Murdoch
To         John Gohde
Subject    Re: Ide Design

  JG> My
 JG> IDE currently has  a limit of about 250K or  about 7,500 lines or
 JG> apparently 4 times the 5.5 file  size limit. That may improved by
 JG> release time,  but I honestly  don't think a  programmer's editor
 JG> needs to deal with really big files. That feature would only take
 JG> up RAM.

But sometimes programmers need to edit big files that aren't programs.  A
programmer's editor doesn't need to be optimized for these, but it's very
handy if it can handle them.

For example, if you were writing a nodelist compiler, and you wanted to stress
test it, you might put some weird errors into a full size version of it.
It's nice to be able to do this in the same editor you're using for everything else.


 JG> The clipboard never made a whole lot of sense to me as it adds an
 JG> extra step.  In my IDE, with  F7/F8 you can move  blocks directly
 JG> from source window  to destination window or you  can implement a
 JG> clipboard approach by using the Delete Buffer (of 300 or whatever
 JG> lines).  8-)

That's a big improvement.  I hate the clipboard.



--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/17  11:00:29

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 09:11:58
From       Dj Murdoch
To         Richard Morris
Subject    Re: Ancestor VMT

  >     (P = Typeof (Self)) or Parent.DescendsFrom(P);

 RM> MMMM<grin>mmmm, you see what I mean!

 >>  If each Object in TP saved it's ancestor's VMT segment in it's
 >> virtual method table then I could query along the chain until I got
 >> to the desired object or a nil VMT segment indicating trhe chain
 >> ended with the root Object Type.  This would extend to the Turbo
 >> vision heirachy.

 > I see the benefits of the feature.  Have you written
 > Borland?

Thinking about this, I see a very big problem both with the code version above
and the modification to the VMT.  It would really interact badly with the
TP linker, as follows:

  Currently, the VMT is only linked in if there's a direct or indirect reference
to it.  TypeOf(Self) makes a direct reference to the VMT; calling a constructor
(without the type modifier prefix) makes an indirect reference to it.  There
aren't any other references that I can think of to the original VMT; all virtual
method calls, for instance, use the pointer in the object, not the fixed VMT
address.

  Same thing for code.  Code is only linked when you give a direct reference
(e.g. call a procedure or static method, or a virtual method with a type overrid
) or an indirect reference.  Indirect references include using procedures
that make direct references, use another routine in the same .OBJ file, or
(and this is the point), make some reference to the VMT.

If you link a type's VMT, you get all the virtual methods, whether you need
them or not.
                                             So what's the point?  Well, think
about that DescendsFrom method or the VMT modification described below it.
 It makes a direct reference to its VMT.  It also makes a direct reference
to Parent.DescendsFrom, which makes a direct reference to the parent VMT,
and so on.  That means that *every* VMT and hence every virtual method for
any ancestor of any object you construct will be linked into your .EXE.

The only way I can see around this problem is to improve the TP linker so
that it takes account of which VMT entries are needed and which aren't.

--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/17  11:00:29

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-16-92 21:29:05
From       Dj Murdoch
To         Trevor Carlsen
Subject    Re: Tp 6.0 Ide

  DW> I have to disagree with you on that one.  The IDE provides at least
 DW> one necessary debugging feature - location of runtime errors in the
 DW> code...

 TC> That is available in any good programming editor - not just the IDE.

Read what he wrote - he's talking run-time errors, not compile errors.  To
get those from other editors you generally have to recompile with the /F option,
and retype the address.


--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/17  11:00:29

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-16-92 21:36:20
From       Dj Murdoch
To         Hans Siemons
Subject    Re: TP 6.0 disassembler...

  GW>> 6.0 It's called TPU60.* (comes to about 120K) and "decompiles" TPU
 GW>> files, including all constants (internal or external), all
 GW>> procedures/functions, objects (untested as yet), optionally provides
 GW>> assembly code (ie. disassembles each procedure) and cross-references
 GW>> everything, and even includes the names of the *.OBJ files (if any)
 GW>> you linked in as external procedures!
 DM> Yes, I've seen that one, and it's really good.  If
 HS> you want to patch a
 DM> .TPU file, that's the one to get.

 HS> Hmm, I want to get it! Do you know where I can freq this
 HS> baby? On 14k4 if possible (I don't mind requesting
 HS> oversees as long as it not on 2400bps ;-)).

I sent it out on the PDN Pascal file echo in October last year as TPU6.ZIP.
 It's probably available closer to you, but the only system I know of for
sure is 1:221/177.  That's 14.4 HST, but I've heard from Europeans that they
have trouble connecting, so watch out that you don't make a lot of wasted
phone calls.

--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/17  11:00:29

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-16-92 21:44:56
From       Dj Murdoch
To         Rob Kittredge
Subject    Re: C++ products

  RK> Terry,
 RK>   I seem to like TP's style better than C's, but I like
 RK> the C++ (3.0) IDE and project management features.  After
 RK> playing with C++'s project management, I can see that (to
 RK> me) it is far superior than specifying the usual 10-15
 RK> USES list in EVERY unit that I write!  Most of which are OP stuff :-).


I though in C++ you generally had to include header files, so you'd end up
with just as many includes as you now have uses.  But even if I'm wrong, it
doesn't sound like a great style.  Units should be focussed - they should
have a clear purpose.  For me, this generally means that they don't get 10
or 15 units in the Uses list - more like 5 or 6.  The main program (or the
unit that dispatches commands to the other units) often ends up with more.


--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/17  11:00:29

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-16-92 21:49:22
From       Dj Murdoch
To         Andrew Knobloch
Subject    Re: Data Compression Techniques ********

  AK>   I program mainly in Turbo PASCAL (with some assembly
 AK> language on the side)... Anyhow, I'd really like to know
 AK> how to compress files and data structures from within a
 AK> PASCAL program.  One algorithm that I do have is called
 AK> Huffman Encoding.  Not only is Huffman Encoding slow but
 AK> it only squishes TEXT files by a maximum of 20 or 30
 AK> percent.  Hardly worth any squishing at all.  Anywayz, if
 AK> anyone knows of a good compression algorithm that is not
 AK> necessarily fast, let me know.

Look for TPLZW.  It was written by Wilbert van Leijen, and does good Lempel-
Ziv-Welch compression pretty quickly.  It won't compress a small file, but
on a long enough text file (e.g. 24K) you get 50 or 60 percent compression.

If you can't find it, hold on for a few days (weeks?).  I'm going to release
 a unit of stream utilities that makes use of Wilbert's code to compress any
stream.

--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/17  11:00:29

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-16-92 21:57:22
From       Dj Murdoch
To         Jud Mccranie
Subject    Exponentiation (was: Are you...)

  JM> Don't hold your breath.  I requested the exponential operator back
 JM> in 2.0 or 3.0.  I even requested in the USCD Apple Pascal days.  They
 JM> called me on the phone and told me how to do it (I knew that, of
 JM> course).  They can write a better one than I can.

Something that's changed since then is the publication and acceptance of the
new Extended Pascal standard.  It has an exponentiation operator ("**", like
Fortran).  I don't know if Borland has any intention of paying attention to
the new standard (any more than they paid attention to the old one :-), but
other vendors on other machines (e.g. VAX Pascal) are.

--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/17  11:00:29

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-16-92 22:01:51
From       Dj Murdoch
To         Ivan Samuelson
Subject    Re: C+ vs Turbo Pascal

  IS> Just C in general is an excellent language. I've got both
 IS> Turbo Pascal and C++. but if you want a language that
 IS> allows you flexibility in manipulating right down to the
 IS> bits, C is the way to go. Pascal is VERY restrictive in
 IS> what you can and cannot manipulate.

Standard Pascal is very restrictive, but I can't think of anything in C that's
impossible in TP 6.  Could you give some examples?  I suspect you just aren't
familiar enough with it.

There are definitely cases where the C construct will be more efficient &
probably shorter to code, but I don't think there's anything that just can't
be done in TP.  (There are also cases where the TP code will be more efficient
and/or shorter.  For example, compile the TurboCalc sample program in both
C & TP, and I think you'll find the TP version is smaller.)

 IS> Pascal was created to be a learning language. It was and
 IS> is used to teach programming techniques in
 IS> structured/modular programming. Hence, it doesn't allow
 IS> you to manipulate on the bit-wise level of computing.
 IS> Also, Pascal will do range checking on arrays. If you step
 IS> out of bounds, Pascal will catch it. C WON'T do that. You
 IS> have to do it yourself. That's where some will see the
 IS> downfall of C while others, like myself, say it's great. C
 IS> gives YOU the responsibility to check everything, which
 IS> allows for greater flexibility in programming.

Now I'm sure it's because of unfamiliarity.  In TP, the $R option controls
whether range checking takes place.  Turn it off if you don't want it.

--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/17  11:00:30

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-17-92 03:33:28
From       Trevor Carlsen
To         Ivan Samuelson
Subject    C+ vs Turbo Pascal



 IS> Just C in general is an excellent language. I've got both Turbo Pascal

 IS> and C++. but if you want a language that allows you flexibility in
 IS> manipulating right down to the bits, C is the way to go. Pascal is
 IS> VERY restrictive in what you can and cannot manipulate.

 IS> Pascal was created to be a learning language. It was and is used to
 IS> teach programming techniques in structured/modular programming. Hence,

 IS> it doesn't allow you to manipulate on the bit-wise level of computing.

 IS> Also, Pascal will do range checking on arrays. If you step out of
 IS> bounds, Pascal will catch it. C WON'T do that. You have to do it
 IS> yourself. That's where some will see the downfall of C while others,

 IS> like myself, say it's great. C gives YOU the responsibility to check

 IS> everything, which allows for greater flexibility in programming.

When comparing languages it is only fair to do so if you know what you are
talking about. The above comments that you make about Pascal are true if you
are referring to standard Pascal.  However your subject line AND the first
quoted line above would indicate that you are referring to, or at the very
least including, Turbo Pascal in your assertations.  If so then your statement
is wrong on several points.

TP *can* easily manipulate bits.
(Operators used for this are shl, shr, and, or, xor, not)

TP *can* disregard range checking of array bounds.
(Read up on the R+/- compiler directive)

TeeCee




--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/17  20:24:20

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-17-92 04:11:45
From       Trevor Carlsen
To         Jan Hoolwerf
Subject    Pascal <==> C string translation



 JH> Probably not a new question but i would appreciate to have some
 JH> routines to translate C strings (null terminated) to Pascal strings.

 JH> And back to C strings. :-)

C strings are just arrays of characters that are terminated by a nul character.
TP strings are arrays of characters that use the zero element of the array
to indicate the length of the string.  Naturally this limits TP strings to
255 characters.

 function Asc2Str(var s; max: byte): string;
   { Converts an ASCIIZ string s to a Turbo Pascal string }
   { with a maximum length of max.                      }
   var starray  : array[1..255] of char absolute s;
       len      : integer;
   begin
     len        := pos(#0,starray)-1;                       { Get the length }

     if (len > max) or (len < 0) then               { length exceeds maximum }

       len      := max;                                  { so set to maximum }

     Asc2Str    := starray;
     Asc2Str[0] := chr(len);                                    { Set length }

   end;  { Asc2Str }

 type
   CStr = array[0..255] of char;

 procedure TPStr2Asc(var s: string; var a: CStr);
   { Converts a TP string s to a nul terminated C string a }
   begin
     a[length(s)] := #0;
     move(s[1],a[0],length(s));
   end;  { TPStr2Asc }

TeeCee



--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/17  20:24:20

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-17-92 06:13:58
From       Trevor Carlsen
To         Gerald Gutierrez
Subject    Printer Status



 > Let's Stamp Out Assembler on Pascal! ;-) Gee, I bet the assembler
 > solution  will save  me .000000043  seconds execution  time in my
 > program.  What a lost.

 GG> And just a COUPLE bytes of exe data.. :) Pascal is fine for small
 GG> applications that don't take much time -- but once it gets up to the

 GG> 50+ line procedures, ASM helps tremendously in both execution time and

 GG> size. IE: Last graphics scroll routine worked on; approximately 150
 GG> lines Pascal code, about 170 lines ASM code, execution times 25+ times

 GG> faster.

I must agree with John Gohde on this.  There is no need to resort to BASM
or assembler code in TP programs unless there is justifiable reason for it.

I define "justifiable reason" as being one of the following -

  * The code is speed sensitive to the point where the use of assembler
    is essential for that speed sensitive routine.
  * Where its use will make a noticeable difference to the end user's
    perception of the speed the program runs at.
  * Where TP is unable to perform the defined task.
  * When size of exe code is critical and use of BASM code reduces the
    code size significantly.
  * When I do not wish to use a particular unit (such as DOS) and BASM
    allows me to easily create a replacement procedure or function from
    that unit.
  * When using assembler is the easiest method to use.

Too often we see code posted here that, using the above criteria, appears
to use BASM unnecessarily.  When I see such code I tend to think the poster
is on an ego trip. :-)

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/17  20:24:20

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-17-92 06:44:09
From       Trevor Carlsen
To         Andreas Holtmann
Subject    Text Files



 AH> const lines = 1024;

 AH> var myfile : array [1..lines] of string;
 AH>     file_handle : text;
 AH>     name = string;
 AH>     line : integer;

This would not even compile, let alone run.  To enable compilation to be complet
d the value of lines would have to be reduced to 253 or less.  (No structure
can exceed just under 64K in size and the total of global variables is limited
to just under 64K.)

TeeCee



--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/17  20:24:20

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-17-92 07:01:25
From       Trevor Carlsen
To         Mike Copeland
Subject    BLOCKWRITE QUESTION.



 MC> ... (The EOF is written only at Close time, only at
 MC> the end of the file.)

TP does not add an end-of-file character unless the programmer specifically
adds it. This is explained in detail in the manual under Read as regards reading
of text files. It would be easy to assume that TP adds an eof character to
files if using the read procedure.  Examination of files created with TP by
an utility such as PCTools will show no eof character present. I do believe
however that some reference to this should be included in the section for
the close procedure.

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/17  20:24:20

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-17-92 16:52:27
From       Trevor Carlsen
To         Greg Williams
Subject    Re: Pascal Vs. C



 GW> Hmm..... some compilers don't require an explicit EXPORT command (to

 GW> declare their interfaced/public routines and variables.  One such
 GW> compiler is the JPI Modula-2...  I don't use EXPORT usually, but it
 GW> SHOULD be supported (as it's part of the language spec (I think?!))...


This message bears no relationship to your subject line and is not related
to the subject matter of this conference.  Please take any further discussion
to netmail.

Trevor Carlsen
Moderator.

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/17  20:24:21

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-17-92 08:06:53
From       Dj Murdoch
To         Kenneth Otto
Subject    Re: Borland pascal 6.1?

  KO> So, is Turbo Pascal 6.1 rumored to be on the way?

I haven't heard anything about TP 6.1, but Borland has said on Compuserve
that they plan to release a high end version of TP called "Borland Pascal"
sometime this year.  It's supposed to support (include?) a DOS extender, so
you can write protected mode programs without Windows; if it's like Borland
C++, it'll also support Windows.

I've also seen somewhere that Borland plans eventually to produce an OS/2
version of TP, but probably not this year.

--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/18  11:26:05

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-17-92 08:10:37
From       Dj Murdoch
To         Mike Copeland
Subject    Re: Pascal, The Language

  JM>Do assemblers have user-defined data types?

 MC>    Certainly! When you consider how Pascal implements
 MC> user-defined data types (it maps the hardware-defined data
 MC> to its data types), Assembler can do the same, and more...

More like "the same, and less".  Most assemblers don't do type checking beyond
the size of the argument; that's a lot less useful than Pascal's strong type
checking.

--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/18  11:26:05

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-17-92 08:14:02
From       Dj Murdoch
To         Steve Gabrilowitz
Subject    Re: Delay patch

  SG> OK, but I still think a better solution is to write your
 SG> own delay routine that is not dependant on the speed of the machine.

I think the point is that that is quite difficult.  If you want it to run
accurately on anything from a 4.77 MHz PC to a 1000 Mhz 786, you'll need to
do some fancy programming.

--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/18  11:26:06

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-17-92 08:16:14
From       Dj Murdoch
To         J J Marquez
Subject    Re: Are you listening... ?

  JJ>    Wanna know what my greatest fear is ? That C++ will add
 JJ> it in first, causing me to have to learn C.... AAAaaaaaaiiiieeeeee!

I'm pretty sure that most numerical libraries for C++ add it in.  There are
some advantages to being able to redefine operators.

--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/18  11:26:06

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-18-92 06:38:44
From       Trevor Carlsen
To         Dale Frakes
Subject    Variable # parameters



 DF> Does anyone out there know how to write a pascal (TP 6.0) function or

 DF> procedure that accepts a variable number of parameters.  I assume this

 DF> is possible, since Write, Writeln, Read, Readln, Concat, etc all
 DF> accept variable number of parameters, and they are written in Pascal.

 DF> Has someone who has seen the source for the Run Time Libraries know
 DF> how this is done?  Again, I would assume that if someone can compile

 DF> the RTL, it can be done within the Pascal environment.

Strictly speaking it cannot be done.  The procedures you refer to that *appear*
to do that are actually changed by the compiler to effectively be *several*
procedures in the actual exe file.

You could however create a procedure that gives the effect of a variable number
of parameters by passing a single parameter which is a linked list.  I've
never found that it is necessary to do this in the time I have been programming.

If you want to emulate the write/writeln/read/readln procedures at some time,
it is easy to set up a TFDD (text file device driver) to do so.  The next
issue of PNL (#11) may have an article explaining this very useful technique.

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/18  18:50:18

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-18-92 13:35:36
From       Trevor Carlsen
To         Frank Livaudais
Subject    Text Files (1 of 3)



 FL> I want to read in text files bt make them binary so that you cant
 FL> 'type' them from dos.

I think that the fastest and easiest way of doing this would be to use some
form of very simple encryption.  Here is a program to encrypt or decrypt a
file - any file - using a simple password as your key and a key file that
you have on your system that is not subject to ever being altered. This approach
s being totally secure as it is almost a "one time pad" method.

It is tested as far as compilation is concerned and also briefly for valid
execution.  It appears to work Ok.

{$A+,B-,D-,E+,F-,G+,I+,L-,N-,O-,R-,S-,V-,X+}
{$M 4048,0,131040}
program encrypt;

{ Author Trevor J Carlsen - released into the public domain 1992         }
{        PO Box 568                                                      }
{        Port Hedland                                                    }
{        Western Australia 6721                                          }
{        Voice [phone]  Data [phone]                      }
{        FidoNet 3:690/644                                               }

{ Syntax: encrypt /p=Password /k=Keyfile /f=File                         }
{ Example -                                                              }
{         encrypt /p=billbloggs /k=c:\command.com /f=p:\prog\anyfile.pas }

{         Password can be any alpha-numeric sequence of AT LEAST four    }
{         characters.                                                    }

{         Keyfile is the full path of any file on the system that this   }
{         program runs on.  This file, preferably a large one, must not  }
{         be subject to changes.  This is critical as it is used as a    }
{         pseudo "one time pad" style key and the slightest change will  }
{         render decryption invalid.                                     }

{         File is the full path of the file to be encrypted or decrypted.}

{ Notes:  Running Encrypt a second time with exactly the same parameters }
{         decrypts an encrypted file.  For total security the keyfile    }
{         can be stored separately on a floppy.  Without this keyfile or }
{         knowledge of its contents it is IMPOSSIBLE to decrypt the      }
{         encrypted file.                                                }

{         Parameters are case insensitive and may be in any order and    }
{         may not contain any dos separator characters.                  }

const
  BufferSize   = 65520;
  Renamed      : boolean = false;

type
  buffer_      = array[0..BufferSize - 1] of byte;
  buffptr      = ^buffer_;
  str80        = string[80];

var
  OldExitProc  : pointer;
  KeyFile,
  OldFile,
  NewFile      : file;
  KeyBuffer,
  Buffer       : buffptr;
  KeyFileSize,
  EncFileSize  : longint;
  Password,
  KFName,
  FFName       : str80;

continued next message.

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/18  18:50:19

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-18-92 13:38:59
From       Trevor Carlsen
To         Frank Livaudais
Subject    Text Files (2 of 3)



... continued from previous message

procedure Hash(p : pointer; numb : byte; var result: longint);
  { When originally called numb must be equal to sizeof    }
  { whatever p is pointing at.  If that is a string numb   }
  { should be equal to length(the_string) and p should be  }
  { ptr(seg(the_string),ofs(the_string)+1)                 }
  var
    temp,
    w    : longint;
    x    : byte;

  begin
    temp := longint(p^);  RandSeed := temp;
    for x := 0 to (numb - 4) do begin
      w := random(maxint) * random(maxint) * random(maxint);
      temp := ((temp shr random(16)) shl random(16)) +
                w + MemL[seg(p^):ofs(p^)+x];
    end;
    result := result xor temp;
  end;  { Hash }

procedure NewExitProc; far;
  { Does the "housekeeping" necessary on program termination }
  var code : integer;
  begin
    ExitProc := OldExitProc;  { Reset exit procedure pointer to original }
    case ExitCode of
    0: writeln('Successfully encrypted or decrypted ',FFName);
    1: begin
         writeln('This program requires 3 parameters -');
         writeln('  /pPassword');
         writeln('  /kKeyFile (full path and name)');
         write  ('  /fFile (The full path and name of the file');
         writeln(' to be processed)');
         writeln;
         write  ('These parameters can be in any order, are case,');
         writeln(' insensitive, and may not contain any spaces.');
       end;
    2: writeln('Could not find key file');
    3: writeln('Could not rename and/or open original file');
    4: writeln('Could not create encrypted file');
    5: writeln('I/O error during processing - could not complete');
    6: writeln('Insufficient memory available');
    7: begin
         writeln('Key  file is too small - aborted');
         writeln;
         writeln(' Key File must be at least as large as the buffer size ');
         write  (' or the size of the file to be encrypted, whatever is the');
         writeln(' smaller.');
       end;
    8: writeln('Password must consist of at least 4 characters');
    else { any other error }
      writeln('Aborted with error ',ExitCode);
    end; { case }
    if Renamed and (ExitCode <> 0) then
      writeln(#7'WARNING: Original file''s name is now TEMP.$$$');
    {$I-}
    close(KeyFile); Code := IOResult;
    close(NewFile); Code := IOResult;
    close(OldFile); Code := IOResult;
    if ExitCode = 0 then
      Erase(OldFile); Code := IOResult;
    {$I+}
  end; { NewExitProc }


function Str2UpCase(var S: string): string;
  { Converts a string S to upper case.  Valid for English. }
  var
    x : byte;
  begin
    Str2UpCase[0] := S[0];
    for x := 1 to length(S) do
      Str2UpCase[x] := UpCase(S[x]);
  end; { Str2UpCase }

continued in next message...





--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/18  18:50:19

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-18-92 13:42:39
From       Trevor Carlsen
To         Frank Livaudais
Subject    Text Files (3 of 3)



... continued from previous message

procedure Initialise;
  var
    CommandLine : string;
    FPos,FLen,
    KPos,KLen,
    PPos,PLen   : byte;

  procedure  AllocateMemory(var p: buffptr; size: longint);
    begin
      if size < BufferSize then begin
        if MaxAvail < size then halt(6);
        GetMem(p,size);
      end
      else begin
        if MaxAvail < BufferSize then halt(6);
        new(p);
      end;
    end; { AllocateMemory }

  begin
    FillChar(OldExitProc,404,0);       { Initialise all global variables }
    FillChar(Password,243,32);
    ExitProc    := @NewExitProc;             { Set up new exit procedure }
    if ParamCount <> 3 then halt(1);
    CommandLine := string(ptr(PrefixSeg,$80)^)+' '; { Add trailing space }
    CommandLine := Str2UpCase(CommandLine);      { Convert to upper case }
    PPos        := pos('/P=',CommandLine);     { Find password parameter }
    KPos        := pos('/K=',CommandLine);      { Find keyfile parameter }
    FPos        := pos('/F=',CommandLine); { Find filename for encryption}
    if (PPos = 0) or (KPos = 0) or (FPos = 0) then    { Parameters wrong }
      Halt(1);
    FFName      := copy(CommandLine,FPos+3,80);
    FFName[0]   := chr(pos(' ',FFName)-1);       { Correct string length }
    KFName      := copy(CommandLine,KPos+3,80);
    KFName[0]   := chr(pos(' ',KFName)-1);
    Password    := copy(CommandLine,PPos+3,80);
    Password[0] := chr(pos(' ',Password)-1);
    if length(Password) < 4 then halt(8);
    { Create a random seed value based on the password }
    Hash(ptr(seg(Password),ofs(Password)+1),length(Password),RandSeed);
    assign(OldFile,FFName);
    {$I-}
    rename(OldFile,'TEMP.$$$');        { Rename the file to be encrypted }
    if IOResult <> 0 then halt(3) else renamed := true;
    assign(OldFile,'TEMP.$$$');
    reset(OldFile,1);
    if IOResult <> 0 then halt(3);
    assign(NewFile,FFName);
    rewrite(NewFile,1);
    if IOResult <> 0 then halt(4);
    assign(KeyFile,KFName);
    reset(KeyFile,1);
    if IOResult <> 0 then halt(2);
    EncFileSize := FileSize(OldFile);
    KeyFileSize := FileSize(KeyFile);
    if KeyFileSize > EncFileSize then KeyFileSize := EncFileSize;
    if IOResult <> 0 then halt(5);
    {$I+}
    if (KeyFileSize < BufferSize) and (KeyFileSize < EncFileSize) then
    halt(7);
    AllocateMemory(buffer,EncFileSize);
    AllocateMemory(KeyBuffer,KeyFileSize);
  end; { Initialise }

procedure Main;
  var
    BytesRead : word;
    finished  : boolean;

  procedure CodeBuffer(number: word);
    { This is the actual encryption/decryption engine }
    var x : word;
    begin
      for x := 0 to number - 1 do
        buffer^[x] := buffer^[x] xor KeyBuffer^[x] xor Random(256);
    end; { CodeBuffer }

  begin
    {$I-}
    finished := false;
    repeat
      BlockRead(OldFile,buffer^,BufferSize,BytesRead);
      if (FilePos(KeyFile) + BytesRead) > KeyFileSize then
        seek(KeyFile,0);
      BlockRead(KeyFile,KeyBuffer^,BytesRead,BytesRead);
      CodeBuffer(BytesRead);
      finished := BytesRead < BufferSize;
      BlockWrite(NewFile,buffer^,BytesRead);
      if IOResult <> then halt(5);
    until finished;
    {$I+}
  end;  { Main }

begin
  Initialise;
  Main;
end.

TeeCee


--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/18  18:50:19

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-15-92 23:34:00
From       Terry Hughes
To         Joshua Kersey @ 930/22
Subject    ExecSwap


JK@9>I need help using ExecSwap, can anyone help me out? This
JK@9>little source does work, it executes it, but ARJ has no
JK@9>memory to work with at all, what's catching my shirttail
JK@9>here?

JK@9>In the ExecWSwp DOcs it says to use FreePtr^, but in TP,
JK@9>it's an "Unknown Identifier" and haven't stooped to buy
JK@P>TPRO yet...

FreePtr has nothing to do with TPRO. FreePtr is a system
variable used in the heap managment in versions 4 through 5.5 of
Turbo Pascal.

The EXECSWAP documentation explains three possible purposes for
LastToSave: 1) save no heap, 2) save all heap but free list and
3) save all heap and free list. The third condition is the one
that requires the expression you were trying to use:
Ptr(Seg(FreePtr^)+$1000, 0). Under Turbo Pascal 6.0, which you
apparently are using, this expression is no longer necessary
since the free list is stored within the heap itself. So, case 2
is no longer applicable and case 3 becomes simply "HeapPtr".

In case that wasn't clear, just pass HeapPtr for LastToSave and
all will be well:

 if InitExecSwap(HeapPtr, 'EXECDEMO.$$$') then
   ...

-Terry

 * OLX 2.2 * TurboPower Software (voice 719-260-6641)

--- Maximus 2.01wb
 * Origin: The Programmers Playhouse (1:128/60)
 * Tossed by SFToss v1.00b on 92/03/19  04:34:20

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-17-92 15:46:02
From       Terry Hughes
To         Rob Kittredge
Subject    C++ products


RK> What was it that convinced you that TP was "better"
RK> than C++ in particular?

Just about everything. <g>

I can't compare the environments since I don't use either IDE.

To the languages themselves, TP is a much leaner/cleaner
language than C++. Bucking the trend here, I hope Borland keeps
TP simple and avoids the temptation to add all the myriad
features (gee gaws?) of C++ to TP. To me, TP's simplicity is
what makes it such a pleasure to work with and what can make a
TP programmer more productive than a C++ programmer.

Beyond that, the biggest issues were compilation capacity,
compilation speed, linking speed and ease of smart linking (in
C++, smart linking requires putting every smart-linkable
procedure in its own OBJ file). TP wins hands down in all of
those categories (although BC++ 3.0 caught up in compilation
capacity).

RK>  P.S. did you happen to take a look at my mods to OPRO for
RK>TP's MAKEHELP to support those "+1" constants?

We decided not to merge those changes into the production
MAKEHELP since it's not a request we've ever heard before.
However, we've kept your version of MAKEHELP around should the
need ever arise. Thanks for sending it.

-Terry

 * OLX 2.2 * TurboPower Software (voice 719-260-6641)

--- Maximus 2.01wb
 * Origin: The Programmers Playhouse (1:128/60)
 * Tossed by SFToss v1.00b on 92/03/19  04:34:20

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-17-92 15:47:04
From       Terry Hughes
To         Joshua Kersey @ 930/22
Subject    Re: Execute With Swap


JK@9>No, it wasn't posted, but the code's kinda long, I'll post
JK@9>it, if it's agreeable to Terry Hughes, it's part ASM and
JK@9>part Pascal..

It's OK by me (since it's PD code) but instead why don't I just
mention our BBS number here (719-260-9726). That way, Kevin can
come get it if he wants it and we'll avoid generating the
lengthy messages that the code would require.

-Terry


 * OLX 2.2 * TurboPower Software (voice 719-260-6641)

--- Maximus 2.01wb
 * Origin: The Programmers Playhouse (1:128/60)
 * Tossed by SFToss v1.00b on 92/03/19  04:34:21

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-18-92 08:39:06
From       Terry Hughes
To         Jud Mccranie
Subject    B-tree filer


JM>First, a slight misprint in the BTree manual, page 64, row 12,
JM>"NextKey" should be "BTNextKey".

Thanks. We'll fix that in the next printing.

JM>up for editing.  Then if I press pageDown (for next record), I use
JM>the NextPrev proc, but the first (only) time I do that it always calls
JM>up the very first record with the matching key, which is not generally
JM>the previous record.

BTNextKey has two distinct behaviors. If the internal
"sequential pointer" is valid then BTNextKey ignores the passed
parameters and returns the next key in sequence. If the internal
sequential pointer is invalid then BTNextKey uses the passed
parameters to reposition the sequential pointer. Note that this
second behavior only occurs if the SearchForSequential option
has been turned on (as it is by default).

Based on your description, I would guess that
SearchForSequential is on and the internal sequential pointer
is invalid or was never set to begin with. A quick way to test
this theory would be to turn SearchForSequential off -- in that
case your call to NextPrevRecord should return 10255 when trying
to get the next record and 10265 when trying to get the previous
record.

You might want to look at the reference section for BTNextKey
for a more complete discussion of this issue.

-Terry

 * OLX 2.2 * TurboPower Software (voice 719-260-6641)

--- Maximus 2.01wb
 * Origin: The Programmers Playhouse (1:128/60)
 * Tossed by SFToss v1.00b on 92/03/19  04:34:21

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-18-92 08:47:08
From       Terry Hughes
To         David G. Edwards
Subject    Why I Like Tp 5.5 Ide (c


DGE>In a message of <10 Mar 92 13:54:43>, Stephan Patterson (1:240/507.3)
writes

DGE> >Like it or not,
DGE> >SAA/CUA-compliant user interfaces are the way of the future. Most people
DGE> >prefer them, by far, to the older style interface a la OPRO.

DGE>I thought OPRO could be used to design SAA/CUA user-interfaces.  Am I off?


You are correct. OPRO can be used to build CUA compliant
interfaces by virtue of its flexibility. And OPRO has had
specific support for dialog boxes (with list boxes, radio
buttons, check boxes, etc.) since version 1.03.

-Terry

 * OLX 2.2 * TurboPower Software (voice 719-260-6641)

--- Maximus 2.01wb
 * Origin: The Programmers Playhouse (1:128/60)
 * Tossed by SFToss v1.00b on 92/03/19  04:34:21

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-18-92 09:03:10
From       Terry Hughes
To         Dj Murdoch
Subject    Re: Binary Tree Linked L


DM> > Actually B-Trees are more effecient than Binary tress and
DM> DF>
DM> DF> I thought the B in B-tree stood for Binary.  What is the
DM> DF> difference between a b-tree and a binary tree?  (Just a
DM> DF> basic explanation, I don't want source and such).

DM>B is for Balanced - to the extent possible, every node has
DM>the same number of subnodes.  A binary tree can be so
DM>imbalanced as to be nearly a linked list, if new nodes are
DM>always added on the right, for instance.

In the circles that I travel (database toolbox vendors and
programmers) B-Tree is generally understood to mean Bayer-Tree
-- Bayer being one of the inventors of this particular
structure. It's different from a binary tree in that each node
contains pointers to many subnodes, rather than just left/right
subnodes and every node must be at least half-full. I believe a
Bayer-Tree implies balancing as well but I don't have my
reference here right now to check.

-Terry

 * OLX 2.2 * TurboPower Software (voice 719-260-6641)

--- Maximus 2.01wb
 * Origin: The Programmers Playhouse (1:128/60)
 * Tossed by SFToss v1.00b on 92/03/19  04:34:22

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-17-92 23:34:10
From       Dj Murdoch
To         All
Subject    Yet another use for streams

 My streams unit keeps growing.  The latest addition is a suite of procedures
that let you use any stream for your overlay buffer.  You say

  OvrInit('whatever.ovr');
  OvrInitStream(AStream);

and it will copy the various overlay segments out to the stream, so that future
overlay loads come from there.  If you have an XMS stream, you can have your
overlays buffered in XMS.  If you have a ramdisk, you can buffer just your
overlays there, without having to copy the whole file over before you run it.


One nice feature is that you can mix and match:  if you have just a bit of
EMS, you can use it for a few overlays, put some more in XMS, a few more on
a ramdisk, and leave the rest on your regular disk.  You can change the allocati
n dynamically, too.

Streams are a lot of fun.

--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/19  12:33:59

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-19-92 06:14:55
From       Trevor Carlsen
To         Dj Murdoch
Subject    Re: Tp 6.0 Ide



 TC> That is available in any good programming editor - not just the
 TC> IDE.
 DM>
 DM> Read what he wrote - he's talking run-time errors, not compile errors.

 DM>  To get those from other editors you generally have to recompile with

 DM> the /F option, and retype the address.

So was I. (Referring to RUN-time errors.)

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/20  04:32:24

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-19-92 06:27:46
From       Trevor Carlsen
To         Gary Elfert
Subject    Re: Reading from a text f



 GE> Could you post an example of how to search for certain text within a
 GE> text file?  Thanks.

Download PNL010.* from a BBS near you.  It contains an example of a Boyer-Moore
search.

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/20  04:32:24

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-19-92 06:33:45
From       Trevor Carlsen
To         Mark Dixon
Subject    Re: Hacking the Network



Take this thread somewhere else.  The subject of this conference is Pascal.

Trevor Carlsen
Moderator

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/20  04:32:25

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-19-92 06:52:09
From       Trevor Carlsen
To         Tim Bogel
Subject    virus



In a message in the PASCAL conference 15 March 92
and addressed to All, Tim Bogel (1:120/320) writes:

 TB>    Does anyone know how to make a virus through pascal, if it is even

 TB> possible, and if so how can I go about making one?

Anyone responding positively to this request can expect to be excommunicated
from this conference with no warning given.  Bogel (and his sysop) have been
netmailed.

Trevor Carlsen
Moderator.



--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/20  04:32:25

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-19-92 07:08:38
From       Trevor Carlsen
To         ANDREW BROUGHTON
Subject    Re: Variable-Length Records



 AB> The database already exists, and will be updated maybe only once a
 AB> year or so, so the updating procedure does not have to be fast or
 AB> careful. The main problem is, the database is about 4 megs, containing

 AB> more than 250,000 SINGLE FIELD records. Now to create an index file is

 AB> the usual way, but the index file would be as big as the database
 AB> file!...

Perhaps you misunderstand the function of the suggested index file.  All the
index file needs to do is give you the offset of each record in the file.
 As the file is about 4MB, the index file can be as small as 3 bytes per record.
(4 bytes if you don't feel like writing routines for a 3 byte integer type.)
Therefore it need not exceed 750K for 250K records and whilst that is still
a 18% increase in storage space requirements, it may well be worth it.

 AB> ... The original database is already sorted...

An index file is certainly the fastest and IMHO the best way to solve the
problem.

You provide no indication of what information the data base contains other
than it being single field.  If you choose a 4 byte index record the 10 msbits
would be available for some kind of information that might further reduce
the need to acces the main file.

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/20  04:32:25

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-19-92 07:27:58
From       Trevor Carlsen
To         Mark Farnan
Subject    Btrieve Programmers



 MF> Not yet, but I would like to learn about it, Have you or anybody else

 MF> got some good source/reference material regarding FAST database
 MF> programming ??, my problem is i need a database where a record is NOT

 MF> a set length, and it also may grow but up to 200 bytes per record at

 MF> anytime.  I DO  NOT want to allcocate this wasted space when i start

 MF> because 200 bytes x 4000 records is ALOT of room, and only some
 MF> records may grow, etc, etc.

Have you got the B-File package from Turbo Power?

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/20  04:32:25

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-19-92 07:29:35
From       Trevor Carlsen
To         Jud Mccranie
Subject    Re: Why I Like Tp 5.5 Ide More Than 6.0



 JM> I sometimes use $IFDEF test, but usually I want to test at that
 JM> location only temporarily.  I need to remember to remove it.

Why?  That is the whole point of compiler conditionals!  They allow you to
leave these extras there without them affecting your executable by just changing
one define right at the start of your code.

Even if you wish to remove them at some later time they make things easier.

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/20  04:32:26

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-19-92 07:33:58
From       Trevor Carlsen
To         Jud Mccranie
Subject    Re: hiya



 JM> randomize doesn't produce a more random number - it just keeps it
 JM> from generating the same sequence each time.

I realise that the following might be construed as taking pedanticism to extreme
, and I'm sure you know what Randomize does, but it doesn't really produce
a different sequence of numbers at all, just ensures a different starting
point in a "master" sequence which is reasonably unpredictable due to the
way it seeds RandSeed *providing* it is not repeatedly called within the same
clock tick.

Now the reason for my pendantics is because some may have interpreted your
statement to mean that use of the Randomize procedure will ensure a seemingly
different entry point for the sequence of numbers each time it used.  Because
of the manner in which it obtains the value it assigns RandSeed, this is only
true if different calls do not occur within the same clock tick.  It also
means there are only 1,573,040 possible starting numbers each day even though
Borland's algorithm has a much longer "period".

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/20  04:32:26

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-19-92 18:23:39
From       Trevor Carlsen
To         Jim Starke
Subject    Input Routine (1 of 3)



 JS> The only thing I can think of is making a totally new input
 JS> routine from scratch that will check for the back space, cursor
 JS> keys, home key, insert key, end key, etc. and re-invent the
 JS> wheel in other words.

 JS> Can anyone enlighten me with some code so I don't have to
 JS> re-invent the wheel?

Here is a unit that should do what you want...

unit keyinput;

{ Author Trevor J Carlsen - released into the public domain 1991         }
{        PO Box 568                                                      }
{        Port Hedland                                                    }
{        Western Australia 6721                                          }
{        Voice [phone]  Data [phone]                       }
{        FidoNet 3:690/644                                               }

{ This unit is designed to permit controlled input into a pre-determined }
{ field size.  It also provides some handy associated procedures and     }
{ functions and constants.                                               }

interface

uses crt;

const
  F1  = $3b00; ShF1  = $5400; CtrlF1  = $5e00; AltF1  = $6800;
  F2  = $3c00; ShF2  = $5500; CtrlF2  = $5f00; AltF2  = $6900;
  F3  = $3d00; ShF3  = $5600; CtrlF3  = $6000; AltF3  = $6a00;
  F4  = $3e00; ShF4  = $5700; CtrlF4  = $6100; AltF4  = $6b00;
  F5  = $3f00; ShF5  = $5800; CtrlF5  = $6200; AltF5  = $6c00;
  F6  = $4000; ShF6  = $5900; CtrlF6  = $6300; AltF6  = $6d00;
  F7  = $4100; ShF7  = $5a00; CtrlF7  = $6400; AltF7  = $6e00;
  F8  = $4200; ShF8  = $5b00; CtrlF8  = $6500; AltF8  = $6f00;
  F9  = $4300; ShF9  = $5c00; CtrlF9  = $6600; AltF9  = $7000;
  F10 = $4400; ShF10 = $5d00; CtrlF10 = $6700; AltF10 = $7100;

  BackSpace    = $0e08; CtrlBackSpace = $0e7f;
  Tab          = $0f09; Tab_left      = $0f00;
  Enter        = $1c0d; CtrlEnter     = $1c0a;
  InsertKey    = $5200; DeleteKey     = $5300;
  Home         = $4700; CtrlHome      = $7700;
  Endkey       = $4f00; CtrlEnd       = $7500;
  PageUp       = $4900; CtrlPageUp    = $8400;
  PageDn       = $5100; CtrlPageDown  = $7600;
  UpArrow      = $4800; DownArrow     = $5000;
  LeftArrow    = $4b00; CtrlLeftArrow = $7300;
  RightArrow   = $4d00; CtrlRightArrow= $7400;
  Escape       = $011b;

type
  CursorState  = (Off, On, Normal, Block);

const
  InsertOn : boolean = false;

procedure Beep(freq,len: word);
function  CursorStatus: CursorState;
procedure Cursor(Action: CursorState);
procedure NormalCursor;
procedure HiddenCursor;
procedure BlockCursor;
function  ReadStr(width   : word;
                  prompt  : string;
                  attr    : byte;
                  s       : string) : string;

implementation

var
  OriginalStatus : CursorState;
  OldExitProc    : pointer;

procedure Beep(freq,len : word);
  { Beeps the speaker for len thousandths of a second }
  begin
    Sound(freq);
    delay(len);
    NoSound;
  end;  { Beep }

function CursorStatus: CursorState;
  var
    bottom: byte absolute $40:$60;
    top   : byte absolute $40:$61;
    x     : shortint;
  begin
    x     := bottom - top;
    if x < 0 then
      CursorStatus := Off
    else if x = 1 then
      CursorStatus := Normal
    else if x > 1 then
      CursorStatus := Block
    else CursorStatus := On;
  end;  { CursorStatus }


Continued next message...




--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/20  04:32:26

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-19-92 18:21:05
From       Trevor Carlsen
To         Jim Starke
Subject    Input Routine (2 of 3)



... continued from previous message.

procedure Cursor(Action : CursorState);
  { Turn the cursor on/off or make it a block}

  procedure ChangeCursor(top,bottom : byte);
    begin
      asm
        mov ah, $01
        mov ch, top
        mov cl, bottom
        int $10
      end;
    end; { ChangeCursor}

begin
  case action of
    On     : if LastMode = Mono then
                ChangeCursor($0C,$0C)
              else
                ChangeCursor($06,$06);
    Normal : if LastMode = Mono then
               ChangeCursor($0B,$0C)
             else
               ChangeCursor($06,$07);
    Off    : ChangeCursor($20,$00);
    Block  : if LastMode = Mono then
               ChangeCursor($02,$0C)
             else
               ChangeCursor($02,$07);
  end; { case}
end;  { ChangeCursor}

procedure NormalCursor;
  begin
    Cursor(On);
  end; { NormalCursor }

procedure HiddenCursor;
  begin
    Cursor(Off);
  end; { HiddenCursor }

procedure BlockCursor;
  begin
    Cursor(Block);
  end;  { BlockCursor }

function KeyWord : word; assembler;
  { Returns a word value where the msb is the scan code of a keypress    }
  { and the lsb is the asciiz value of the key.                          }
  asm
    mov  ax,0
    int  16h
  end;  { KeyWord }


Continued next message ...




--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/20  04:32:27

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-19-92 18:22:33
From       Trevor Carlsen
To         Jim Starke
Subject    Input Routine  (3 of 3)



...continued from previous message.


function  ReadStr(width   : word;
                  prompt  : string;
                  attr    : byte;
                  s       : string) : string;

{   Editing keys are -                                                   }
{     DeleteKey        - DeleteKeys character at the cursor.             }
{     LeftArrow        - Nondestructive move cursor to the left.         }
{     RightArrow       - Nondestructive move cursor to the right.        }
{     End              - Move cursor to end of input string.             }
{     Home             - Move cursor to start of input string.           }
{     Backspace        - DeleteKeys character to the left of cursor.     }
{     escape           - Aborts routine which then returns just an       }
{                        escape character - regardless of what           }
{                        characters were previously entered.             }
{     return/enter     - Leaves routine with string returned.            }

{   Width = The width of the input field.  Once input reaches the width  }
{           required, no further characters are accepted.                }
{   prompt= A prompt will be displayed in the current attribute.  If no  }
{           prompt is required pass a nul string.                        }
{   attr  = The input field will be displayed in attr colour.            }
{   s     = s will be displayed in the input field and the cursor will   }
{           be positioned at the end of the s string.                    }

{   Example:                                                             }
{      st := ReadStr(20,'Enter Name: ',LtGrayOnBlue,st);                 }
{      ( st MUST be initialised in the above example before the call. )  }

const
    space = #32;
  var
    xpos, ypos,
    stpos,OldAttr : byte;
    len           : byte absolute s;
    finished      : boolean;
    key           : word;
    ch            : char absolute key;

  procedure WriteField;
    var x : byte;
    begin
      GotoXY(xpos,ypos);
      for x := 1 to width do
        write(space);
      GotoXY(xpos,ypos);
    end; { WriteField }

  procedure DeleteChar;
    begin
      Delete(s,stpos,1);
      s := s + space;
      gotoXY(xpos,ypos);
      write(s);
      dec(len);
    end;  { DeleteChar }

  procedure AddChar;
    { Checks that it is valid to insert or add a character to input str }
    begin
      if InsertOn then begin
        if (len < width) then begin
          move(s[stpos],s[succ(stpos)],width-pred(stpos));
          inc(len);
          s[stpos] := ch;
          inc(stpos);
        end
        else beep(450,15);
      end else begin
        if stpos <= width then begin
          s[stpos] := ch;
          if stpos > len then
            inc(len);
          inc(stpos);
        end else beep(450,15);
      end;
    end; { AddChar }

  begin
    finished      := false;
    OldAttr       := TextAttr;
    write(prompt);
    TextAttr      := attr;
    xpos          := WhereX;
    ypos          := WhereY;
    WriteField;
    stpos         := succ(len);
    repeat
      GotoXY(xpos,ypos);
      write(s);
      GotoXY(xpos + pred(stpos),ypos);
      if InsertOn then
        Cursor(Block)
      else
        Cursor(Normal);
      key := KeyWord;
      case key of
        InsertKey  : InsertOn := not InsertOn;
        DeleteKey  : if (len > 0) and (stpos > 0) then
                       DeleteChar;
        Enter      : begin
                       ReadStr  := s;
                       finished := true;
                     end;
        BackSpace  : if stpos > 1 then begin
                       dec(stpos);
                       DeleteChar;
                     end
                     else beep(450,15);
        Escape     : begin
                       finished := true;
                       ReadStr  := ch;
                       WriteField;
                       len      := 0;
                     end;
        LeftArrow  : if stpos > 1 then dec(stpos);
        RightArrow : if stpos <= len then inc(stpos);
        Home       : stpos := 1;
        EndKey     : stpos := succ(len);
        else begin case lo(key) of
                     32..255: AddChar
                     else beep(450,15);
                   end; { case lo(key) of }
             end; { else }
      end; { case key of }
    until finished;
    TextAttr      := OldAttr;
    writeln;
  end; { ReadStr }

procedure KbdExitProc; far;
  begin
    ExitProc := OldExitProc;
    Cursor(OriginalStatus);
  end;  { KbdExitProc }

begin
  { Set up an exit procedure to ensure that the cursor is restored }
  OldExitProc    := ExitProc;
  OriginalStatus := CursorStatus;
end.



--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/20  04:32:28

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-19-92 19:37:33
From       Trevor Carlsen
To         James Chambers
Subject    Text Viewer (1 of 3)



 JC> Does anyone know where I can get the source for a
 JC> simple text viewer?
 JC> I just need to display one 66 line page on screen
 JC> and allow scrolling.

I'm not sure how you intend to display 66 lines on screen.  However here is
a text viewer that will allow you to view a text file with scrolling capabilitie
.

{$A+,B-,D-,E+,F-,I+,L-,N-,O-,R+,S-,V-}
{$M 4048,65536,655360}

program Readtext;

{ Author Trevor J Carlsen - released into the public domain 1991         }
{        PO Box 568                                                      }
{        Port Hedland                                                    }
{        Western Australia 6721                                          }
{        Voice [phone]  Data [phone]                       }
{        FidoNet 3:690/644                                               }

{ This example programs displays a text file using simple word wrap. The }
{ cursor keys are used to page forward or backwards by page or by line.  }
{ The program makes some important assumptions.  The main one is that no }
{ line in the file will ever exceed 255 characters in length.  To get    }
{ around this restriction the ReadTxtLine function would need to be      }
{ rewritten.                                                             }

{ The other major restriction is that files exceeding a size able to be  }
{ totally placed in RAM cannot be viewed.                                }

{.$DEFINE TurboPower (Remove the period if you have Turbo Power's TPro)  }

uses
  {$IFDEF TurboPower }
  tpcrt,
  colordef;
  {$ELSE}
  crt;
  {$ENDIF}

const
  {$IFNDEF TurboPower }
  BlackOnLtGray = $70;      LtGrayOnBlue = $17;
  {$ENDIF}
  LineLength    = 79;       MaxLines     = 6000;
  ScreenLines   = 22;       escape       = $011b;
  Home          = $4700;    _end         = $4f00;
  upArrow       = $4800;    downArrow    = $5000;
  PageUp        = $4900;    PageDown     = $5100;

type
  LineStr    = string[Linelength];
  StrPtr     = ^LineStr;

var
  TxtFile    : text;
  Lines      : array[1..MaxLines] of StrPtr;
  NumberLines: 1..MaxLines+1;
  CurrentLine: 1..MaxLines+1-ScreenLines;
  st         : string;
  finished   : boolean;
  OldExitProc: pointer;
  TxtBuffer  : array[0..16383] of byte;

function LastPos(ch : char; S : string): byte;
  { Returns the last position of ch in S or zero if ch not in S }
  var
    x   : word;
    len : byte absolute S;
  begin
    x := succ(len);
    repeat
      dec(x);
    until (x = 0) or (S[x] = ch);
    LastPos := x;
  end;  { LastPos }


continued in next message ...


--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/20  04:32:28

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-19-92 19:41:56
From       Trevor Carlsen
To         James Chambers
Subject    Text Viewer (2 of 3)



... continued from previous message.

function Wrap(var S,CarryOver: string): string;
  { Returns a string of maximum length Linelength from S. Any additional }
  { characters remaining are placed into CarryOver.                      }
  const
    space = #32;
  var
    temp      : string;
    LastSpace : byte;
    len       : byte absolute S;
  begin
    FillChar(temp,sizeof(temp),32);
    temp := S; CarryOver := ''; wrap := temp;
    if length(temp) > LineLength then begin
      LastSpace := LastPos(space,copy(temp,1,LineLength+1));
      if LastSpace <> 0 then begin
        Wrap[0]   := chr(LastSpace - 1);
        CarryOver := copy(temp,LastSpace + 1, 255)
      end  { if LastSpace... }
      else begin
        Wrap[0]   := chr(len);
        CarryOver := copy(temp,len,255);
      end; { else }
    end; { if (length(S))...}
  end;  { Wrap }

function ReadTxtLine(var f: text; L: byte): string;
  var
    temp : string;
    len  : byte absolute temp;
    done : boolean;
  begin
    len := 0; done := false;
    {$I-}
    while not eoln(f) do begin
      read(f,temp);
      if IOResult <> 0 then begin
        writeln('Error reading file - aborted');
        halt;
      end;
    end; { while }
    if eoln(f) then readln(f);
    ReadTxtLine := st + Wrap(temp,st);
    finished := eof(f);
  end;  { ReadTxtLine }

procedure ReadTxtFile(var f: text);
  var
    x : word;
  begin
    st          := '';
    NumberLines := 1;
    repeat
      if NumberLines > MaxLines then begin
        writeln('File too big');
        halt;
      end;
      if (MaxAvail >= Sizeof(LineStr)) then
        new(Lines[NumberLines])
      else begin
        writeln('Insufficient memory');
        halt;
      end;
      FillChar(Lines[NumberLines]^,LineLength+1,32);
      if length(st) > LineLength then
        Lines[NumberLines]^  := wrap(st,st)
      else if length(st) <> 0 then begin
        Lines[NumberLines]^  := st;
        st := '';
      end else
        Lines[NumberLines]^  := ReadTxtLine(f,LineLength+1);
      Lines[NumberLines]^[0] := chr(LineLength);
      if not finished then
        inc(NumberLines);
    until finished;
  end;  { ReadTxtFile }

procedure DisplayScreen(line: word);
  var
    x : byte;
  begin
    GotoXY(1,1);
    for x := 1 to ScreenLines - 1 do
      writeln(Lines[x-1+line]^);
    write(Lines[x+line]^)
  end;

procedure PreviousPage;
  begin
    if CurrentLine > ScreenLines then
      dec(CurrentLine,ScreenLines-1)
    else
      CurrentLine := 1;
  end;  { PreviousPage }

continued in next message ...

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/20  04:32:29

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-19-92 19:44:03
From       Trevor Carlsen
To         James Chambers
Subject    Text Viewer (3 of 3)



... continued from previous message.
procedure NextPage;
  begin
    if CurrentLine < (succ(NumberLines) - ScreenLines * 2) then
      inc(CurrentLine,ScreenLines-1)
    else CurrentLine := succ(NumberLines) - ScreenLines;
  end;   { NextPage }

procedure PreviousLine;
  begin
    if CurrentLine > 1 then dec(CurrentLine)
    else CurrentLine := 1;
  end;  { PreviousLine }

procedure NextLine;
  begin
    if CurrentLine < (succ(NumberLines) - ScreenLines) then
      inc(CurrentLine)
    else CurrentLine := succ(NumberLines) - ScreenLines;
  end; { NextLine }

procedure StartOfFile;
  begin
    CurrentLine := 1;
  end; { StartOfFile }

procedure EndOfFile;
  begin
    CurrentLine := succ(NumberLines) - ScreenLines;
  end;  { EndOfFile }

procedure DisplayFile;

  function KeyWord : word; assembler;
    asm
      mov  ah,0
      int  16h
    end;

  begin
    DisplayScreen(CurrentLine);
    repeat
      case KeyWord of
        PageUp    : PreviousPage;
        PageDown  : NextPage;
        UpArrow   : PreviousLine;
        DownArrow : NextLine;
        Home      : StartOfFile;
        _End      : EndOfFile;
        Escape    : halt;
      end; { case }
      DisplayScreen(CurrentLine);
    until false;
  end; { DisplayFile }

{$IFDEF TurboPower}
procedure NewExitProc;far;
  begin
    ExitProc := OldExitProc;
    NormalCursor;
  end;
{$ENDIF}

procedure Initialise;
  begin
    CurrentLine := 1;
    if ParamCount <> 1 then begin
      writeln('No file name parameter');
      halt;
    end;
    assign(TxtFile,Paramstr(1));
    {$I-}  reset(TxtFile);
    if IOResult <> 0 then begin
      writeln('Unable to open ',Paramstr(1));
      halt;
    end;
    SetTextBuf(TxtFile,TxtBuffer);
    Window(1,23,80,25);
    TextAttr := BlackOnLtGray;
    clrscr;
    writeln
('              Next Page = [PageDown]     Previous Page = [PageUp]');
    writeln
('              Next Line = [DownArrow]    Previous Line = [UpArrow]');
    write
('         Start of File = [Home]   End of File = [End]   Quit = [Escape]');
    Window(1,1,80,22);
    TextAttr := LtGrayOnBlue;
    clrscr;
    {$IFDEF TurboPower}
    HiddenCursor;
    OldExitProc := ExitProc;
    ExitProc    := @NewExitProc;
    {$ENDIF}
  end;

begin
  Initialise;
  ReadTxtFile(TxtFile);
  DisplayFile;
end.


--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/20  04:32:29

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-18-92 19:47:00
From       Norbert Igl
To         Chris Elris
Subject    my bed





 > SL> tied to the bed and lick you all over.  i then would move down to
 > SL> and suck it hard.....
 > great babe, your bed or mine?!?!

  ??????  ..... hmmmmm  (:-))

---
 * Origin: Let's talk about hex, baby.... (2:241/5300.3)
 * Tossed by SFToss v1.00b on 92/03/20  11:33:04

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-18-92 17:28:17
From       Dj Murdoch
To         Jud Mccranie
Subject    Re: Why I Like Tp 5.5 Ide (continued)

  JM> What if you had to think about which pedal to press each time while
 JM> driving?  That would be a mess.  But with the 6.0 IDE you have to
 JM> think about it each time, unlike the previous versions.

Come on, that's just nonsense.  *You* have to think about it, because you're
new to the 6.0 IDE.  Anyone who uses it much learns the new keys just the
way you learned the old Wordstar keys.


--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/20  11:34:20

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-18-92 17:31:14
From       Dj Murdoch
To         Mike Wilson
Subject    Re: Runtime Error 200 On An Xt.

  MW> Okay, what would you use for a do all catch all {$g+}
 MW> grep?  I think I have tried the -l but I have tried
 MW> everything else I can find and this still seems to be a problem...

I'd use:

  grep -r- -i+ "g+" *.*

and probably turn up a fair bit of junk, but would be sure to turn up what
I was looking for if it was there.

--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/20  11:34:20

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-18-92 17:32:34
From       Dj Murdoch
To         David G. Edwards
Subject    Re: Pascal Structure Philosophy

  DG> That brings up something I'd like to see in a future
 DG> version of TP: I'd like to look at the compiler's output
 DG> like the CPU window in TD.

 You know about my DUMPPROG utility?  It does a dump something like the TD
window.  I'm thinking about a new release; just looking for a 486-aware disassem
ler first.  Once I get that, I'll finish off some work I've done on resolving
numeric addresses as symbolic names.

--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/20  11:34:20

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-18-92 20:21:18
From       Dj Murdoch
To         Dale Frakes
Subject    Re: Variable # parameters

  DF> Does anyone out there know how to write a pascal (TP 6.0)
 DF> function or procedure that accepts a variable number of
 DF> parameters.  I assume this is possible, since Write,
 DF> Writeln, Read, Readln, Concat, etc all accept variable
 DF> number of parameters, and they are written in Pascal.  Has
 DF> someone who has seen the source for the Run Time Libraries
 DF> know how this is done?  Again, I would assume that if
 DF> someone can compile the RTL, it can be done within the Pascal environment.

No, you can't do it even if you have the RTL, unless you want to call your
procedure Writeln, Read, Readln, Concat, etc., and lose access to the real
one.  Those names are hard coded into a part of the RTL for which you don't
get source (it's not Pascal, it's probably produced by another program), and
the compiler has special code for those particular functions/procedures and
no others.

One solution is the one used by TurboVision:  have your function work on lists
and produce a pointer to a list, so that you can just add parentheses to your
heart's content (see the TV manual for examples).  I think a better solution
is to just live with the fact that the number of parameters is fixed, and
design around it.

What particular use do you have in mind?


--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/20  11:34:20

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-18-92 20:31:22
From       Dj Murdoch
To         Chris Long
Subject    Re: Obsolete pascal??

  CL> -> Hi everyone!  I would like to know if the Pascal Language is going to

 CL> -> become obsolute in the next 5 years!

 CL> Nonsense.  The language is powerful, flexible, easy to
 CL> learn, and fun to program in.  Those are winning
 CL> attributes...  Of course, the industry has adopted C as
 CL> its preferred language, so C will undoubtedly be more
 CL> popular than pascal....  But, that will not make the language obsolete.

I'd say that Standard Pascal is close to obsolete right now.  I would never
use it, nor would I recommend that anyone else use it.  Turbo Pascal, on the
other hand, is a useful language that will probably still be popular in 5
years.  It's not compliant with the Pascal standard, though, so I wouldn't
call it Pascal.

It's possible that Extended Pascal will be around in 5 years.  I don't know
if it will catch on or not; I tend to doubt it, but hope otherwise.  Its big
problem is that Turbo Pascal isn't compatible with it either, so it doesn't
really stand a chance on the PC unless Borland chooses to adopt it.


--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/20  11:34:21

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-19-92 08:20:53
From       Dj Murdoch
To         ANDREW BROUGHTON
Subject    Re: Variable-Length Records

  AB> Thanks for replying. I thought of that, but my situation
 AB> is unique. The database already exists, and will be
 AB> updated maybe only once a year or so, so the updating
 AB> procedure does not have to be fast or careful. The main
 AB> problem is, the database is about 4 megs, containing more
 AB> than 250,000 SINGLE FIELD records. Now to create an index
 AB> file is the usual way, but the index file would be as big
 AB> as the database file!

Don't put keys into the index file, just put the offset of the start of the
corresponding record.  Because of the size, you'll need to use longint offsets;
that means the index will be about 1 Meg.  It might be worth your while to
do some fiddling and create 3 byte integers; that would reduce the size of
the index to 750K.

 AB> So far, I can only see 2 ways
 AB> to deal with this: One is to copy the database over, and
 AB> make each record the same length, (and probably doubling
 AB> the size), or just do the binary search as if the file
 AB> were made up of records all of the same size, and when the
 AB> seek is done, increment or decrement the file pointer
 AB> until it is on the start of a record, then check it, then
 AB> continue the search.

The second method would work well, if you could easily recognize the start
of a record.  If they contain binary data, that won't be easy.  If they're
just text strings, and you're sure certain characters will never occur (e.g.
 high ascii, or control characters) you could easily construct a signature
to put at the start of a record that couldn't possibly occur anywhere else.



--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/20  11:34:21

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-19-92 08:30:55
From       Dj Murdoch
To         Jud Mccranie
Subject    Re: TP 6 Bad design

  JM> I ran them from EXE files.  To be sure, I did it again.

 JM> TP   286   seconds

 JM> 5.5  N/A   22.24
 JM> 6.0  G-    22.90
 JM> 6.0  G+    22.96 (even though the EXE was 1.3% smaller than above)

Did you recompile before re-running?  A significant part of that difference
could just be load time; if the 5.5 .EXE is contiguous on the disk, near the
FAT, it'll load noticeably faster than an .EXE that's badly fragmented.

Another possibility is that your disk cache may have the 5.5 code already
in it (because it's on the disk near something that you've just used).

Timing .EXEs isn't easy.  I'd only trust the timings if you've been running
them from a ramdisk.

--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/20  11:34:21

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-19-92 08:35:43
From       Dj Murdoch
To         Joshua Kersey @ 930/22
Subject    Re: Are you listening... ?

 JJM>    If you don't do scientific or engineering calculations...  JJM> there
is very little in it for you. What I am asking for is a JJM> type REAL exponenti
tion operator (NOT a function, but an JJM> OPERATOR, like + or * ) which can
raise a real number to a real JJM> (non-integer) power.

 JK> Ahhh...I seee, yes, I've been needing one too, and yes,
 JK> I'll agree! "**" Does sound like a very good operator to
 JK> use too...good idea. ..

But JJ isn't ambitious enough.  There's no problem writing a function to raise
a real number to a real power.  The hard thing is writing an operator that
raises integers to integer powers, reals to integer powers, or reals to real
powers, all with the same syntax, and all choosing efficient ways to do the
calculation.

For example,

   Pi**2.0

should be calculated as Exp(2.0*ln(Pi)), but

  Pi**2

should be calculated as Pi*Pi, and

  2**10

should be calculated as 1 shl 10.

(Actually, all of the above should be calculated by the compiler, and the
value should be put into the expression; but the calculations I give are the
method to use if the constants were variables.)


--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/20  11:34:22

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-20-92 04:49:32
From       Trevor Carlsen
To         Daniel Walton
Subject    Re: BLOCKWRITE QUESTION.



 DW>  I haven't had a chance to test it yet, but I read a
 DW> paragraph in "Mastering Turbo Pascal 6"  by Scott D. Palmer
 DW> that says otherwise. It says, " One potential pitfall, if
 DW> you do use Blockwrite, is that it inserts an end-of-file
 DW> character (Ctrl-Z) at the end of a block. This means that if
 DW> a block ends in the middle of an existing file and you look
 DW> for the end of the file by using the EOF function, your
 DW> program will stop at the end of the block." Am I just
 DW> reading this wrong or what?

I would suggest that, if that is a direct quote, you are reading it correctly.
 The trouble is it is very wrong!  If the man can make a statement that is
a stuff up of that magnitude, it does not auger well for the quality of the
rest of the book.

I think "Mastering Turbo Pascal 6" is also the title of a book by Tom Swan
which has received rave reviews here.  I have not seen it but other books
by him are of excellent standard.

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/21  11:38:09

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-20-92 05:15:07
From       Trevor Carlsen
To         Andrew Broughton
Subject    Variable-Length Records



 AB> I have a database that is about 4.5 megs, containing over
 AB> 250,000 one-field entries. The database is already sorted in
 AB> alphabetical order...

Mmmm... I think that (sorted in alpha) is the little bit of vital information
previously not given :-). (You previously stated that it was sorted but not
how it was sorted.)

 AB> An index file would be too big, and a memory variable would
 AB> be impossible. The conclusion that I've come to so far is to
 AB> treat the file as if it contained equal-length records, then
 AB> start the binary search. If the file pointer lands in the
 AB> middle of a record, then Inc or Dec it until it gets to the
 AB> beginning of the record.

Here is an idea that would make your search *very* much easier and faster.
 It would also involve an index file of fixed size (52728 bytes).

First create a "mediumint" type! (Necessary so that the array will fit in
one segment)

type
  MediumInt = array[0..2] of byte;

Create an array['A'..'Z','A'..'Z','A'..'Z'] of MediumInt, then read each record
of the file and store the offset of each record whose first three characters
differ from the previous records first three characters in this array.  For
the example we'll call this array - "Offsets".

Then to find the first name SMITH, all that is required is to -
  seek(f,Medium2Long(Offsets[name[1],name[2],name[3]]);

You can then either sequentially search all records from that offset on until
the appropriate one is found or not found (by reaching SMJ) or you can BlockRead
the whole file between SMI and SMJ and do a Boyer-Moore search of the buffer.
 The latter method would mean you would find the record you want in under
1/10th second if all records starting "SMI" were less than one segment in
size.

If you want code for using a "mediumint" I will post it.

TeeCee



--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/21  11:38:09

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-20-92 06:06:38
From       Trevor Carlsen
To         Andrew Broughton
Subject    Variable-Length Records



My previous message to you contains a small error.  In the line that reads -

"...read each record of the file and store the offset of each record..."
                                                         ^^^^
Change that to -
"...read each record of the file and store the offset of the first record..."
                                                         ^^^^^^^^^

Sorry 'bout that!

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/21  11:38:10

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-20-92 08:04:55
From       Trevor Carlsen
To         Steve Sparks
Subject    Re: Hacking the Network



Final WARNING:

This thread has no place in this conference.  Take it elsewhere immediately.

If you, or anybody else, wish to reply to this message do so by netmail only.

Trevor Carlsen
Moderator.

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/21  11:38:11

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-20-92 08:19:02
From       Trevor Carlsen
To         Ashley Nichols
Subject    Pascal 6.0



 AN> How do you write a linked list to disk? And how can you tell
 AN> the differnce between the different ones? I can insert
 AN> things into the list, but yet I can't tell the difference
 AN> between them. I don't know anything about linked lists, and
 AN> so I would like all the help I can get!

Obviously the pointer parts of a linked disk mean nothing if written to disk.
So just traverse the list writing the relevant parts of each record to a disk
record.

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/21  11:38:11

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-20-92 08:21:20
From       Trevor Carlsen
To         Gene Buckle
Subject    Re: Bbs Source.. / Range Checking



 GB> The variable is being set to false as the first command in
 GB> the main begin..end clause.  I'm not terribly concerned
 GB> about it right now as I don't have the time to work on the
 GB> program anymore.  Thanks for the info.  Might save me some
 GB> time on a later project.

OK.  So the rcs variable is somehow getting trashed. Watch it in the debugger.

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/21  11:38:11

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-20-92 17:41:09
From       Trevor Carlsen
To         Jud Mccranie
Subject    Re: Pascal, The Language



 TC> difficulties of using assembler.  The power of a language is
 TC> not how easy it is for a programmer to achieve something in
 TC> that language but how efficient and capable the language is in
 TC> executing that something.

 JM> Again you're equating execution speed and small size to
 JM> power...

No way! Execution speed and size have no real relationship to the power of
a language.  They are directly related to the efficiency of a language.

 JM> ...The power of the language is its power in expressing what
 JM> we need it to.

Exactly - or perhaps instead of the second "power" you could say "capability".
Either way you put it better than I did.  If a language "A" cannot do something
at all and language "B" can do that something, regardless of how easy or hard
it is to achieve that, and language "B" can also do everything that language
"A" can, then language "B" is obviously the more powerful. That is the case
with TP and assembler.  Execution wise, assembler can do everything TP can
but TP cannot do everything that assembler can.  There are some things that
simply cannot be done in TP, no matter how adept you are with the language.
Therefore assembler is more powerful than TP.  As a bonus it is also more
efficient (faster) and produces much smaller executable code and makes things
safer and easier for the programmer with its strong type checking, user defined
types etc.

Your definition of power appears to be how easy a language makes programming
a particular task.  I think I would call that its usability.

 JM> If I wanted a program to run 3 times as fast I could buy a
 JM> computer 3 times as fast as what I have now for $3000.  To
 JM> code everything in assembler to make it 3x faster would take
 JM> programmer-centuries.

At no time would I consider suggesting that someone produces entire applications
in assembler.  That would often be tantamount to economical suicide.  All
I am suggesting is that those selected parts of a program that a user may
perceive to be slow should be critically examined - probably with the help
of a profiler - and an intelligent decision made as to whether there would
be any user benefit in recoding small critical parts in assembler.  Often
just a few lines of assembler can result in massive time savings.

 JM> In my big program, printing takes a long time, unless a
 JM> print spooler is used.  Otherwise, the longest it takes for
 JM> anything is about 3 seconds to read in the files at the
 JM> beginning.  Everything else that the user does takes less
 JM> than a second.  Coding in assembler isn't going to speed up
 JM> the i/o appreciably.  Everything else is fast enough as far
 JM> as the user is concerned.  Cutting that 0.5 second to 0.2
 JM> second by using assembler would take much more effort than
 JM> it is worth.

Fine, and I would usually agree with you there.  What I am trying to say however
 is that by not learning or using assembler, you are severely crippling yourself
by denying yourself the tools for that inevitable time when some assembler
is needed and is essential.  Make no mistake, as a professional programmer
that time *will* come, if you are not willing to take advantage of every opportu
ity, you may as well throw money down the gurgler.

Just so there is no misunderstanding I reiterate, I rarely use assembler,
in fact I avoid it like the plague, I am not extra good at it, but ...,  I
would never say I don't use it or have no need of it.  I'm too fond of the
folding stuff to do that! :-)

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/21  11:38:12

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-20-92 08:55:48
From       Trevor Carlsen
To         Jud Mccranie
Subject    Re: Tp 6.0 Ide



 TC> In any testing a fundamental rule involves testing the limits,
 TC> upper and lower, of all data.  This involves stack limits, heap
 TC> limits, the whole works.  In the IDE it would be difficult, if
 TC> not impossible to do this properly.  TSRs and ISRs cannot
 TC> really be tested at all from the IDE unless you are fond of
 TC> rebooting!

 JM> OK, I don't know about TSRs and ISRs since I've never
 JM> written one. They must be different than app progs.  As far
 JM> as the stack and heap limits, they are larger when running
 JM> the EXE, so if you don't hit the limit in the IDE the EXE
 JM> won't either.

You have either misunderstood or chosen to misinterpret what I have said.
It is *essential* in testing any application that you *do* hit the design
limits.  This is really the only way you can be sure that your program will
behave as you want it too when it is working to absolute maximum.  It is not
much use to sell a program that crashes when some limit is reached because
this behaviour was never tested before release.

TeeCee



--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/21  11:38:12

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-20-92 16:53:41
From       Trevor Carlsen
To         Jud Mccranie
Subject    messages



cc: Greg Smith (all versions)

 JM> I got a message from one Greg Smith telling me to quit
 JM> sending him messages about Pascal.  But I only replied to
 JM> messages on the Pascal echo from a Greg Smith.  Is Greg #2
 JM> being confused by messages to Greg #1?

Almost certainly.  The name is probably not that uncommon.  We also have two
Greg Smiths here in Australia on the network. There is bound to be more than
just the two you have discovered in the US, one of whom is well liked and
well respected in this echo.  The complaint shows a real lack of BBS "sense"
and awareness of the scope of these conferences.  Treat it with the contempt
it deserves and ignore it.

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/21  11:38:12

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-20-92 16:53:20
From       Trevor Carlsen
To         Greg Smith
Subject    (All GS versions!)



Original to Jud McCranie in reply to the following quote...

 JM> I got a message from one Greg Smith telling me to quit
 JM> sending him messages about Pascal.  But I only replied to
 JM> messages on the Pascal echo from a Greg Smith.  Is Greg #2
 JM> being confused by messages to Greg #1?

Almost certainly.  The name is obviously not all that uncommon.  We also have
two Greg Smiths here in Australia on the network. There is bound to be more
than just the two you have discovered in the US, at least one of whom is well
liked and well respected in this echo.  The complaint shows a real lack of
BBS "sense" and awareness of the scope of these conferences.  Treat it with
the contempt it deserves and ignore it.

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/21  11:38:12

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-20-92 09:12:23
From       Trevor Carlsen
To         Jud Mccranie
Subject    Re: C++ products



 JM> I haven't done any C (except for a couple of trivial progs),
 JM> but I agree with you!  In fact, I've just read several
 JM> messages advising beginners to "learn Pascal first then move
 JM> up to C".  Why not learn Pascal and stick with it????  I
 JM> guess it depends on two issues: portability and the level of
 JM> the app.

I reckon there is a third important one - the age and aspirations of the learner
  If, like me, s/he is a little older than most and has no ambitions to ever
become a professional programmer in the sense of entering that job market,
then it is of little consequence to learn C, as TP will do everything C does,
often faster and more efficiently.

However, if the learner is young and does wish to make himself more marketable
in the job market, then a thorough grounding in as many languages as possible,
*particularly* C, is indicated.

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/21  11:38:13

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-20-92 10:20:13
From       Trevor Carlsen
To         Jud Mccranie
Subject    Re: TP 6 Bad design



 JM> I ran them from EXE files.  To be sure, I did it again.

 JM> TP   286   seconds

 JM> 5.5  N/A   22.24
 JM> 6.0  G-    22.90
 JM> 6.0  G+    22.96 (even though the EXE was 1.3% smaller
 JM> than above)

 JM> Of course, the 0.06 seconds isn't significant, it may be
 JM> when the tick occured.  I have R-, A+, S-, I- in the source.
 JM> One more thing, I have {$M 65520, 0, 6553600} which (in tp
 JM> 5.5) makes it 4% faster than not having the line.  I haven't
 JM> tested w/ and w/o that line in 6.0.  A+ is 2% faster, S- is
 JM> 8% faster, in my old tests.

Two things here surprise me.  The first is that 5.5 is faster and the second
is that the memory directive has so much effect.

Could you write a short demo that illustrates this first point and post it
here?  As for the second point, I cannot reproduce that nor think of why it
should be.  I'll be interested to see the code.

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/21  11:38:13

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-20-92 10:28:19
From       Trevor Carlsen
To         Gavin Campbell
Subject    Screen Save



 GC> X:    FillChar(screensave, SizeOf(screensave));

 GC>   Well I had a couple problems compiling this.  First on
 GC> line six it said '=' expected.  So i fixed that.  But on the
 GC> fillchar command it said ',' needed.  I dont know how to fix
 GC> that. other than that its just what i wanted.

It may surprise you, but the answer is contained on page 34 of the TP6 manual
and page 268 of the TP5.5 manual.  Another surprise may be that it comes under
the explanation for the procedure FillChar.  Try reading it.

Another way you could help yourself is to (from the IDE) place the cursor
on FillChar and press Ctrl-F1 ...

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/21  11:38:13

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-20-92 17:08:30
From       Trevor Carlsen
To         Gerald Gutierrez
Subject    Echo policemen



 GG>  1. Leave moderation to the moderator. Self appointed "echo policemen"
 GG>     only waste echo space and create ill-feeling.
 GG>     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

 GG> It's greatly appreciated if you could follow this one .. it saves
 GG> a lot of unnecessary messages.

Mmm... reminds me of T S Eliot's quote - "You! hypocrite lecteur! -mon semblable
- mon frere!"  which he got from Charles Baudelaire! :-)

TeeCee

--- TC-ED   v2.01
 * Origin: The Pilbara's Pascal Centre ([phone]) (3:690/644)
 * Tossed by SFToss v1.00b on 92/03/21  11:38:13

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-20-92 12:14:00
From       Werner Berghofer
To         Jud McCranie
Subject    Tpascal mags.




       Jud,

 > Could two "Greg Smith"s cause this problem?

       obviously there are at least two users called Greg Smith active in
this echo: the one Greg Smith from 1:104/441.8 and the other Greg Smith from
1:110/69.  However, only the first Greg Smith seems to be interested in Pascal.

       W.

---
 * Origin: God is real... unless declared integer. (2:310/90.100)
 * Tossed by SFToss v1.00b on 92/03/21  11:38:41

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-19-92 23:23:27
From       Dj Murdoch
To         David G. Edwards
Subject    Stream errors

 I noticed you asked your question about DOS errors in streams on Compuserve
again, and it made me curiouser.  I have the TPW source, and I think the stream
code there is very close (identical?) to the stream code in TV.  In all the
cases I checked, with the possible exception of GetSize (where something weird
was going on), the Error method gets called immediately after the DOS call
that fails.  It has Info set to the value returned by DOS in AX, which is
the error code.  It looks to me as though it would always be safe to ask DOS
for extended error information at that point.

You could probably also ask for it in the StreamError global procedure, but
there's a chance that some stream's Error method would already have messed
with it before you got a chance to see it.  If you're writing this for your
own use, I'd do it that way, and make a note to myself to be careful when
overriding Error.

--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/21  11:40:27

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-20-92 08:48:45
From       Dj Murdoch
To         Jud Mccranie
Subject    Re: TP 6.0 IDE memory

  DM> It's probably too late to make any difference now, but what
 DM> sort of changes were you looking for?

 JM> Back to 5.5-style IDE, obviously.

No, you misunderstood my question.  What sort of improvements on 5.5 were
you looking for?

--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/21  11:40:27

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-20-92 08:49:44
From       Dj Murdoch
To         Dale Frakes
Subject    Re: Turbo Vision Questions...

  > And apart from the TStream class and its descendants, and the
 > TStringList, and the TResourceFile.  All of which are useful
 > in programs that have nothing descended from TView.

 DF> Aren't TStream, TStringList, and TResourceFile all
 DF> decendants of TObject?  I thought the only object in TV
 DF> that isn't a decendant of TObject is TPoint (and TRect).
 DF> Am I thinking about something in the wrong way?  Is this
 DF> why I'm having trouble with Turbo Vision?

You're right.  The point is that you *can* pick and choose from the TObject
hierarchy; it's just the TView sub-hierarchy that you need to take or leave
as a whole.

--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/21  11:40:28

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-20-92 08:53:22
From       Dj Murdoch
To         Jud Mccranie
Subject    Re: Exponentiation (was: Are you...)

  JM> Great.  Did I also hear that they're going to add complex?

I think type complex was in the standard.  I don't know if anyone is going
to add it to a real compiler, but I'd guess VAX Pascal.  They seem to be the
big standard supporters.

--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/21  11:40:28

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-20-92 08:55:52
From       Dj Murdoch
To         Terry Hughes
Subject    Re: Binary Tree Linked L

  TH> In the circles that I travel (database toolbox vendors and
 TH> programmers) B-Tree is generally understood to mean Bayer-Tree
 TH> -- Bayer being one of the inventors of this particular
 TH> structure. It's different from a binary tree in that each node
 TH> contains pointers to many subnodes, rather than just left/right
 TH> subnodes and every node must be at least half-full. I believe a
 TH> Bayer-Tree implies balancing as well but I don't have my
 TH> reference here right now to check.

Thanks.  I think I've heard that before, but I forgot it.  Why don't they
just call them Bayer-Trees?

--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/21  11:40:28

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-20-92 18:19:30
From       Dj Murdoch
To         Jud Mccranie
Subject    Re: Obsolete Pascal??

  JM> Interestingly, though, BASIC and FORTRAN have gone through such
 JM> radical changes that old versions won't even come close to compiling.
 JM> Pascal isn't that way, indicating that it was designed better.

Well, standard Pascal won't compile in TP, whereas Fortran 66 compiles fine
in Fortran 77 and probably in the latest standard as well.  I think Fortran
is a much worse design than Pascal, but it has a lot of staying power.


--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/21  11:40:28

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-20-92 20:58:23
From       Dj Murdoch
To         Jud Mccranie
Subject    Re: How to group procs in units

  JM> How do you think it is best to organize your units?  That is, should
 JM> you

 JM> 1. group procs into units according to functionality (e.g. all
 JM>    file handling together, all reports together, all editing
 JM>    together, etc)

 JM> 2. group according to level (ones that don't call any other procs
 JM>    together, all procs that call only ones in the previous unit,
 JM>    etc)

 JM> 3. Some combination of these two

 JM> 4. Something else

Most of my programs use two quite different kinds of units.  There are the
high level ones, that are specific to each program, and the low level service
units, which are shared among many programs.  The organization is different
for the two kinds.

The high level units should be organized according to the organization of
the program.  If a program does three "things", then there should be three
high level units, and perhaps a fourth which dispatches requests to the other
three.  These units will handle all the high level functionality involved
in their particular task, and will call on the low level units to actually
carry out the tasks.  If more than one task creates reports, then I may well
put the report generating capability separately into each of them.  On the
other hand, if the reports are closely related, and should have the same appeara
ce, I'll make an intermediate level report writing unit, and the two high
level units will both make calls to it.

The low level units should be organized strictly according to functionality.
 If I want to do something with a mouse, I'd expect to find it in the Mouse
unit.  Something with strings should be in the Strings unit.  (In actual fact,
the names for these are OPMouse and OPString :-).

The common thing tying both approaches together is the rule:  Every unit in
a program should have a one line description.  If I can't describe what a
unit does in one line, then there's something wrong with the way that unit
is designed.  The different lines should be distinctive, too - it shouldn't
be hard to guess where a particular function/procedure/variable would be,
just from scanning a list of the one-liners.  If it could be in either of
two units, then they should probably be merged, or at least rearranged, so
that the distinction between them is clearer.

In fact, the rule above is just a special case of the general rule:  *everything
 in a program should have a one line description.  If I can't summarize what
something is in a single line, then it means I don't have a clear idea, and
it needs work.

--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/21  11:40:29

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Conference 4
Date       03-20-92 21:16:36
From       Dj Murdoch
To         Daniel Walton
Subject    Re: BLOCKWRITE QUESTION.

  DW>  I haven't had a chance to test it yet, but I read a
 DW> paragraph in "Mastering
 DW> Turbo Pascal 6"  by Scott D. Palmer that says otherwise. It says, " One

 DW> potential pitfall, if you do use Blockwrite, is that it
 DW> inserts an end-of-file
 DW> character (Ctrl-Z) at the end of a block. This means that if a block ends

 DW> in the middle of an existing file and you look for the end
 DW> of the file by using
 DW> the EOF function, your program will stop at the end of the block."
 DW> Am I just reading this wrong or what?

If that's an exact quote, then it's nonsense.  TP 6 doesn't put a Ctrl-Z in
when it closes a file, and certainly doesn't put one in at the end of a block
written by BlockWrite.

--- Msg V3.2
 * Origin: Murdoch's Point: Waterloo, Ontario, Canada (1:221/177.40)
 * Tossed by SFToss v1.00b on 92/03/21  11:40:29