Celestin Apprentice 5/Apprentice-Release5.iso/Information/CSMP Digest/volume 3/csmp-digest-v3-111

Text File | 1995-12-31 | 54KB | 1,345 lines

C.S.M.P. Digest             Tue, 19 Sep 95       Volume 3 : Issue 111

Today's Topics:

        Difference between BlockMove() and BlockMoveData()?
        Easiest way to get vRefNum of startup volume?
        Pointers to functions on the PowerPC?
        Preferences Database project discussion
        QuickTime volumes
        Random Numbers
        [Q] Gestalt to determine number of serial ports?



The Comp.Sys.Mac.Programmer Digest is moderated by Francois Pottier
([e-mail]).

The digest is a collection of article threads from the internet newsgroups
comp.sys.mac.programmer.help, csmp.tools and csmp.misc. It is designed for
people who read news semi-regularly and want an archive of the discussions.
If you don't know what a newsgroup is, you probably don't have access to
it. Ask your systems administrator(s) for details. If you don't have access
to news, you may still be able to post messages to the group by using a
mail server like anon.penet.fi (mail [e-mail] for more
information).

Each issue of the digest contains one or more sets of articles (called
threads), with each set corresponding to a 'discussion' of a particular
subject.  The articles are not edited; all articles included in this digest
are in their original posted form (as received by our news server at
nef.ens.fr).  Article threads are not added to the digest until the last
article added to the thread is at least two weeks old (this is to ensure that
the thread is dead before adding it to the digest).  Article threads that
consist of only one message are generally not included in the digest.

The digest is officially distributed by two means, by email and ftp.

If you want to receive the digest by mail, send email to [e-mail]
with no subject and one of the following commands as body:
    help                                Sends you a summary of commands
    subscribe csmp-digest Your Name     Adds you to the mailing list
    signoff csmp-digest                 Removes you from the list
Once you have subscribed, you will automatically receive each new
issue as it is created.

The official ftp info is //ftp.dartmouth.edu/pub/csmp-digest.
Questions related to the ftp site should be directed to
[e-mail].

-------------------------------------------------------

>From [e-mail] (Jeffrey Posnick)
Subject: Difference between BlockMove() and BlockMoveData()?
Date: Fri, 1 Sep 1995 19:42:03 GMT
Organization: The Dorsai Embassy

The title says it all: is there any difference between BlockMove() and
BlockMoveData()? If so, when should I use one and not the other?Thanks for
any info you could offer, MPTA didn't even seem to know BlockMoveData()
existed.

Jeffrey Posnick | <mailto:[e-mail]> | <mailto:[e-mail]>
The opinions expressed above do not necessarily reflect those of my employer (which is a little strange since I'm self employed).

+++++++++++++++++++++++++++

>From [e-mail] (AndrewWelc)
Date: 1 Sep 1995 18:40:53 -0400
Organization: America Online, Inc. (1-800-827-6364)

> The title says it all: is there any difference between BlockMove() and
> BlockMoveData()? If so, when should I use one and not the other?Thanks
> for any info you could offer, MPTA didn't even seem to know
> BlockMoveData() existed.

Yes -- BlockMove flushes the instruction caches (on most block moves),
while BlockMoveData doesn't.  If you're just moving data around,
BlockMoveData will be a bit faster.

Regards,

Andrew Welch
Thaumaturgist
Ambrosia Software, Inc.

..........

For the latest versions of our software, technical support, and Ambrosia
news, stop by and visit the Ambrosia Software, Inc. support forums:

America Online ---> Keyword: Ambrosia
    CompuServe ---> GO word: Ambrosia
        eWorld --> Shortcut: Ambrosia

+++++++++++++++++++++++++++

>From Richard Wesley <[e-mail]>
Date: 2 Sep 1995 14:18:25 GMT
Organization: Punch Deck Consulting

[e-mail] (Jeffrey Posnick) wrote:
>The title says it all: is there any difference between BlockMove() and
>BlockMoveData()? If so, when should I use one and not the other?Thanks for
>any info you could offer, MPTA didn't even seem to know BlockMoveData()
>existed.
>
>Jeffrey Posnick | <mailto:[e-mail]> | <mailto:[e-mail]>
>The opinions expressed above do not necessarily reflect those of my employer (which is a little strange since I'm self employed).

::BlockMove flushes the instruction cache, ::BlockMoveData does not.  You should
use ::BlockMoveData for data and ::BlockMove for anything else.

- rmgw

http://www.punchdeck.com/hawkfish/PunchDeck.html

- --------------------------------------------------------------------------
Richard Wesley  [e-mail] | "'Hand it round first, and cut it
Punch Deck Consulting [e-mail] |  afterwards.'" - Lewis Carroll,
     Macintosh Software Development    |    "Through the Looking Glass"
- --------------------------------------------------------------------------



+++++++++++++++++++++++++++

>From [e-mail] (Frank Durgin, et al.)
Date: Sat, 02 Sep 1995 16:58:44 -0400
Organization: Swarthmore Visual Perception Lab

In article <[e-mail]>,
[e-mail] (Jeffrey Posnick) wrote:

> The title says it all: is there any difference between BlockMove() and
> BlockMoveData()? If so, when should I use one and not the other?Thanks for
> any info you could offer, MPTA didn't even seem to know BlockMoveData()
> existed.

BlockMoveData() doesn't flush the instruction cache (if any), so it's
*slightly* faster on the '040. On 604 machines with the DRE,
BlockMoveData() can be quite a lot faster since (at least in loops)
flushing the caches can mean the difference between recompiling 68k
instructions as PPC once and then running N-1 chunks of PPC code and
recompiling 68k instructions as PPC N times. On machines with no
instruction cache, the two calls are the same.

In other words, if you are copying data rather than instructions, use
BlockMoveData(). Can't hurt; might help.


Ta,
        -Stephen

+++++++++++++++++++++++++++

>From [e-mail] (Donald Way)
Date: Sun, 03 Sep 1995 18:39:17 -0500
Organization: epix.net

In article <[e-mail]>,
[e-mail] wrote:

> In article <[e-mail]>,
> [e-mail] (Jeffrey Posnick) wrote:
>
> > The title says it all: is there any difference between BlockMove() and
> > BlockMoveData()? If so, when should I use one and not the other?Thanks for
> > any info you could offer, MPTA didn't even seem to know BlockMoveData()
> > existed.
>
> BlockMoveData() doesn't flush the instruction cache (if any), so it's
> *slightly* faster on the '040. On 604 machines with the DRE,
> BlockMoveData() can be quite a lot faster since (at least in loops)
> flushing the caches can mean the difference between recompiling 68k
> instructions as PPC once and then running N-1 chunks of PPC code and
> recompiling 68k instructions as PPC N times. On machines with no
> instruction cache, the two calls are the same.
>
> In other words, if you are copying data rather than instructions, use
> BlockMoveData(). Can't hurt; might help.


Jeez, I guess I gotta start reading the headers more often.  When did they
sneak this one in?  No reference to it ANYWHERE as far as I can tell.
It'd be great if everytime Apple did something like this a little dogcow
would start flashing in my task menu.

BTW, do we need to test for the trap, or is there some minimum system
configuration upon which we can simply *assume* that BlockMoveData is
present?

--
Donald Way
[e-mail]

+++++++++++++++++++++++++++

>From [e-mail] (Steve Kiene)
Date: Mon, 04 Sep 1995 01:30:05 -0700
Organization: MindVision Software

In article <[e-mail]>, [e-mail]
(Donald Way) wrote:

> In article <[e-mail]>,
> [e-mail] wrote:
>
> > In article <[e-mail]>,
> > [e-mail] (Jeffrey Posnick) wrote:
> >
> > > The title says it all: is there any difference between BlockMove() and
> > > BlockMoveData()? If so, when should I use one and not the other?Thanks for
> > > any info you could offer, MPTA didn't even seem to know BlockMoveData()
> > > existed.
> >
> > BlockMoveData() doesn't flush the instruction cache (if any), so it's
> > *slightly* faster on the '040. On 604 machines with the DRE,
> > BlockMoveData() can be quite a lot faster since (at least in loops)
> > flushing the caches can mean the difference between recompiling 68k
> > instructions as PPC once and then running N-1 chunks of PPC code and
> > recompiling 68k instructions as PPC N times. On machines with no
> > instruction cache, the two calls are the same.
> >
> > In other words, if you are copying data rather than instructions, use
> > BlockMoveData(). Can't hurt; might help.
>
>
> Jeez, I guess I gotta start reading the headers more often.  When did they
> sneak this one in?  No reference to it ANYWHERE as far as I can tell.
> It'd be great if everytime Apple did something like this a little dogcow
> would start flashing in my task menu.
>
> BTW, do we need to test for the trap, or is there some minimum system
> configuration upon which we can simply *assume* that BlockMoveData is
> present?


You don't need to test for the presence of the trap. It's simply the
_BlockMove trap with the immediate bit set in the trap word. If a
particular ROM doesn't support it, the bit is just ignored.

Steve Kiene
MindVision Software

---------------------------

>From [e-mail] (Eddy J. Gurney)
Subject: Easiest way to get vRefNum of startup volume?
Date: 22 Aug 1995 22:01:24 GMT
Organization: Hewlett-Packard

I want to call PBHGetVInfo and the two easiest ways appear to be
specifying the vRefNum or the volume index.  Can I make any assumptions
about the volume index for the startup volume (i.e., is volume index
0 always the startup volume?) or can I always assume that the vRefNum
will be -1 for the startup volume?

If not, what is the simplest correct way to determine the vRefNum or
volume index of the startup volume?

Thanks,
Eddy

+++++++++++++++++++++++++++

>From [e-mail] (Ed Anson)
Date: Tue, 22 Aug 1995 23:52:12 -0400
Organization: Tulip Software

In article <41dk3k$[e-mail]>, [e-mail] (Eddy J. Gurney)
wrote:

> I want to call PBHGetVInfo and the two easiest ways appear to be
> specifying the vRefNum or the volume index.  Can I make any assumptions
> about the volume index for the startup volume (i.e., is volume index
> 0 always the startup volume?) or can I always assume that the vRefNum
> will be -1 for the startup volume?
>
> If not, what is the simplest correct way to determine the vRefNum or
> volume index of the startup volume?

The System folder is always on the startup volume.

Use the Folder Manager to locate the System folder, and the resulting file
spec will reveal the startup volume.

- --------------------
Ed Anson            MediaTree: multimedia outline editor & catalog
Tulip Software
Andover, MA 01810   For details, check out my WWW page:
U.S.A.              <http://www.tiac.net/users/tulip/home.html>

+++++++++++++++++++++++++++

>From [e-mail] (AndrewWelc)
Date: 23 Aug 1995 00:00:04 -0400
Organization: America Online, Inc. (1-800-827-6364)

> or can I always assume that the vRefNum will be -1 for the startup
> volume?

Yep, according to Apple, it is always -1.

Regards,

Andrew Welch
Thaumaturgist
Ambrosia Software, Inc.

..........

For the latest versions of our software, technical support, and Ambrosia
news, stop by and visit the Ambrosia Software, Inc. support forums:

America Online ---> Keyword: Ambrosia
    CompuServe ---> GO word: Ambrosia
        eWorld --> Shortcut: Ambrosia

+++++++++++++++++++++++++++

>From [e-mail] (Jump Long)
Date: 3 Sep 1995 00:12:51 -0400
Organization: America Online, Inc. (1-800-827-6364)

[e-mail] (AndrewWelc) replied:

>> or can I always assume that the vRefNum will be -1 for the startup
>> volume?
>
>Yep, according to Apple, it is always -1.

With the current File Manager, it is always -1. That because the boot
volume is mounted first and volume reference numbers are allocated
starting with -1. However, that's not something that is guaranteed (and in
fact, it almost changed once).

Use FindFolder to find the System Folder and then use the vRefNum
returned. That will work no matter what changes in the future.

- Jim Luther


---------------------------

>From [e-mail] (Jonathan R. Seagrave)
Subject: Pointers to functions on the PowerPC?
Date: 3 Sep 1995 19:57:23 GMT
Organization: Northwest Nexus Inc.

Hi, I've been having some trouble porting 68k code to PowerPC code.  I
thought maybe I could get a pointer or two.  First, (I'll focus on a
specific problem -- I doubt I'd get much response if I asked 32767
questions at once ;)  I'm having trouble with action functions.  This
worked fine in 68k but I can't get it to run without a crash with the
PowerPC:

    TrackControl(hControl, localMouse, (void *) ScrollActionProc);

/*
the prototype for ScrollActionProc is:
pascal void ScrollActionProc(ControlHandle hControl, short partCode);
*/

Tracing throught my code with a debugger shows that ScrollActionProc() is
never run instead I get an Illegal Instruction message (apparently this
(void *) ScrollActionProc is going of into the ozone...).

I had insert the (void *) to get it to compile; it wouldn't compile any
other way.  Before the switch from 68k too PowerPC the above line read:

    TrackControl(hControl, localMouse, ScrollActionProc);

I've never had problems with it on the 68k.

Interestingly enough although I've had to typecast other function pointers
as void* I haven't seen errors in any others.  For example, these seem to
work fine:

    SetGrowZone((void *) MyGrowZone);
    DeviceLoop(pWind->visRgn, (void *) GrayFrameDrawingProc, (long)
&hFrameData, 0);

I've also been having problems with this:

    ForEachIconDo(hIcon, svAllAvailableData,  (void *) SaveCustomIcon, 0);

But I can't claim it was working before as I'm writting it for the first
time on the PowerPC.

I suspect I'm Ignorant of some aspect of passing pointers to functions.
Any tips would be appreciated.

jrs     http://www.az.com:/~jrs
--
The other day I saw a bumper sticker which read "Honk if you passed
p-chem."  I honked and the driver flipped me off.  I guess he didn't pass

+++++++++++++++++++++++++++

>From [e-mail] (Daniel Azuma)
Date: Mon, 04 Sep 1995 17:40:22 -0700
Organization: California Institute of Technology, Pasadena

[e-mail] (Jonathan R. Seagrave) wrote:

> Hi, I've been having some trouble porting 68k code to PowerPC code.  I
> thought maybe I could get a pointer or two.  First, (I'll focus on a
> specific problem -- I doubt I'd get much response if I asked 32767
> questions at once ;)  I'm having trouble with action functions.  This
> worked fine in 68k but I can't get it to run without a crash with the
> PowerPC:
>
>     TrackControl(hControl, localMouse, (void *) ScrollActionProc);
>
> /*
> the prototype for ScrollActionProc is:
> pascal void ScrollActionProc(ControlHandle hControl, short partCode);
> */
>
> Tracing throught my code with a debugger shows that ScrollActionProc() is
> never run instead I get an Illegal Instruction message (apparently this
> (void *) ScrollActionProc is going of into the ozone...).
>
> I had insert the (void *) to get it to compile; it wouldn't compile any
> other way.  Before the switch from 68k too PowerPC the above line read:
>
>     TrackControl(hControl, localMouse, ScrollActionProc);
>
> I've never had problems with it on the 68k.

Okay, Jonathan, this is the trick when porting to PowerPC: you have to
deal with mode switching.

Parts of the toolbox are fat (both 68K and PPC) but parts are still
68K-only. This 68K-only code cannot call your PPC-native action function
directly; it gets confused and tries to interpret the PPC instructions as
68K instructions, which results in illegal instruction errors.

To deal with this, you need to specify action procedures as Universal
ProcPtrs (UPPs). The UPP includes the address of the function, plus
information on how to make the mode-switch from 68K to PPC. This is also
the reason the code won't compile unless you typecast to (void *):
TrackControl() is not expecting a function pointer, but a UPP.

To do this, you use a macro to create a UPP out of the address of your
action procedure. For the TrackControl action procedure, the macro is
called NewControlActionProc(), and it returns a UPP of type
ControlActionUPP. This is what needs to be passed to TrackControl. Once
you're done, you use the DisposeRoutineDescriptor() call to release the
UPP.

So your code would look like this:

    ControlActionUPP myScrollActionUPP;

    myScrollActionUPP = NewControlActionProc(ScrollActionProc);
    TrackControl(hControl, localMouse, myScrollActionUPP);

    // ...When done...
    DisposeRoutineDescriptor (myScrollActionUPP);

You can look at the corresponding header files in the Universal Headers to
see what the names of the UPP-creating macros are. There's a different
macro for each kind of callback. NewControlActionProc() is defined in
"Controls.h". A modal dialog filter proc UPP is created by the
NewModalFilterProc() macro, defined in "Dialogs.h".

The above code will also work under 68K. When compiling PPC-native, the
macros create the UPPs that are needed. When compiling 68K-native, they
merely perform the necessary typecasts to make the code compile.
Therefore, it is a good idea to use this technique from now on, in any mac
code that you write.

In general, UPPs need to be used any time where it is possible that 68K
code will call PPC code, or PPC code will call 68K code. For instance, you
will need to use them if part of your program is compiled 68K and part is
PPC. The details on this more advanced usage are discussed in "IM: PowerPC
System Software". Some info can also be gleaned by studying the header
file "MixedMode.h".

> The other day I saw a bumper sticker which read "Honk if you passed
> p-chem."  I honked and the driver flipped me off.  I guess he didn't pass

ROTFL! Heh, maybe I should get a bumper sticker saying "Honk if you passed
thermal physics"...

Dan

                                                                +-+
================================================================| |=======
 \ _____ Daniel Azuma _____ \ "See what love the Father has  +--+ +--+  /
  \ <[e-mail]> \  given us, that we should be  +--+ +--+ /
   \    Caltech CS Student    \  called children of God..."     | |   /
    \  Mac Programming Artist  \            ---1 John 3:1       | |  /
    ============================================================| |===
                                                                | |
                                                                +-+

+++++++++++++++++++++++++++

>From [e-mail] (Ed Anson)
Date: Mon, 04 Sep 1995 22:59:53 -0400
Organization: Tulip Software

In article <[e-mail]>, [e-mail] (Jonathan R.
Seagrave) wrote:

> questions at once ;)  I'm having trouble with action functions.  This
> worked fine in 68k but I can't get it to run without a crash with the
> PowerPC:
>
>     TrackControl(hControl, localMouse, (void *) ScrollActionProc);
>
> /*
> the prototype for ScrollActionProc is:
> pascal void ScrollActionProc(ControlHandle hControl, short partCode);
> */
>
> Tracing throught my code with a debugger shows that ScrollActionProc() is
> never run instead I get an Illegal Instruction message (apparently this
> (void *) ScrollActionProc is going of into the ozone...).
>
> I had insert the (void *) to get it to compile; it wouldn't compile any
> other way.  Before the switch from 68k too PowerPC the above line read:

The compiler was trying to tell you something. Too bad you lied to it and
made it shut up :)

PPC code requires that you supply a UPP instead of a normal procedure
pointer, when setting up call-backs such as these. The current "universal"
headers provide the appropriate declarations and methods to create
appropriate UPPs.

By casting away the type information, you fooled the compiler into
accepting a procedure pointer where a UPP was required. The system's (68K)
code tried to call your (PPC) code and things didn't work out as planned.
The UPP is needed to enable the transition between emulated and native
code.

If you don't know what this is all about, there is an entire volume of NIM
devoted to PPC programming. It is essential reading.

- --------------------
Ed Anson            MediaTree: multimedia outline editor & catalog
Tulip Software
Andover, MA 01810   For details, check out my WWW page:
U.S.A.              <http://www.tiac.net/users/tulip/home.html>

+++++++++++++++++++++++++++

>From [e-mail] (Jordan Zimmerman)
Date: Tue, 05 Sep 1995 17:47:09 -0700
Organization: Altura Software, Inc.

> Hi, I've been having some trouble porting 68k code to PowerPC code.  I
> thought maybe I could get a pointer or two.  First, (I'll focus on a
> specific problem -- I doubt I'd get much response if I asked 32767
> questions at once ;)  I'm having trouble with action functions.  This
> worked fine in 68k but I can't get it to run without a crash with the
> PowerPC:
>
>     TrackControl(hControl, localMouse, (void *) ScrollActionProc);
>

You really should read the Inside Mac volume on PowerPC System Software.
Any callback that goes through the MacOS (as in TrackControl) must be
passed as a RoutineDescriptor so that the MixedMode Manager can handle it.
A full description is beyond what can be written here.

--
Jordan Zimmerman, Altura Software
home page: http://www.altura.com/jordanz/home.html
Coming to you fast as lightning on a 9500/120!

---------------------------

>From [e-mail] (HILSABECK DAVID AL)
Subject: Preferences Database project discussion
Date: 2 Sep 1995 14:02:17 GMT
Organization: Oklahoma State University, Stillwater OK

t seems as though every time I develop a Macintosh application, implementing
preferences seems to become more and more of a pain. There's to be a lot of
discussion on the topic, and Apple has published guidelines for implementing
preferences, but none of this answers the underlying question: How can I add
preferences to my application without the pain of managing a preferences
file?

All of the problems preferences pose could easily be overcome if there
were an automatic method of storing and retrieving data associated with an
application. Windows programmers have had a solution (a very limited
solution) to the preferences problem for some time. With .INI files, a
Windows programmer simply called one of four functions to store and
retrieve values in what amounted to preferences files. Now with Windows95
and Windows NT, Windows programmers have the Registration Database, which is
sort of like preferences on excessive steroids. Even the NeXT operating
system had a database for storing preferences, even if it did become corrupt
from time to time.

The Mac OS has gone without long enough. The time has come to implement the
Macintosh Preferences Database. This is my proposal:

Implement a Preferences Database System that is composed of four parts. The
first part is a shared library (and an equivalent ASLM or 68KCFM library)
that implements the Preferences Manager. This provides an API for an
application to store and retrieve preferences in the Preferences Database.
The second part is the collection of Preferences Database files themselves.
These files reside in the Preferences folder and are completely controlled
by the Preferences Manager. The third part is the Preferences Utility. The
Preferences Utility allows end-user to do some simple management of the
contents of the Preferences Database, as well as repair the Preferences
Database should the need arise. The fourth part is the Developer Preferences
Utility. This utility allows a developer to view and manipulate the contents
of the database for debugging purposes. The two utilities may be packaged as
a single utility, with a hidden switch toggling the developer functionality.

The overall structure of the database would be heirarchial. The
root would contain a list of application signatures. Under its signature,
an application can store a heirarchy of any data necessary to implement its
preferences. The key to retrieve any data item from the database would be a
parent level ID (a long) and the name of the data item (an OSType). The root
level would have a predefined ID. There would be excessively practical
limits imposed on the size (and content) of data stored in the database.

The Preferences Database would consist of a index/sequence set stored in one
main database file, along with extra files to store objects that would pose
storage problems for the main database file. The structure of the main
database file would be a b+tree with a b*tree index set. The sequence
set would be based on an adaptive inverted list structure. Both index and
sequence set would rely on an underlying block structure. All algorithms used
to manipulate the index set, sequence set, and block structure would be
fault tolerant.

The Preferences Manager would simply be a set of functions stored in a shared
library. There will be no system extensions, control panels, or background
applications. The API presented by the Preferences Manager would allow an
application to open the database, store and retrieve data within the database
heirarchy, get information about the contents of the database, and otherwise
manipulate the database. The functionality would be somewhat congruent to the
functionality of the File Manager. When Copland is released, the Preferences
Manager would need to modified to allow controlled, shared, re-entrant, and
pre-emptive access to the database (which can't be done now without at least
turning on file sharing) through a [hopefully] new File Manager API. I have
designed an actual Preferences Manager API, but I don't want to get that
detailed quite yet (but I will later).

The functionality of the Preferences Utilities is still fuzzy. What I would
like to see is a architecture that would allow plug-ins to decode the
contents of custom data items for developers. Or maybe something based on
ResEdit TMPLs. Like I said, this is still fuzzy. I can say that the
Registration Manager/Utility in Windows NT is completely overboard. I want
something that won't scare (or harm) end-users.

Finally, the Preferences Database System would be freely distributable. As
more applications used it, its value to both users and developers would
increase. I've even kicked around the idea of making it portable, since
so much cross-platform development is being done these days (trust me, I
KNOW!).

This should be enough to get the ball rolling. If you understand my
preferences frustration and find the Preferences Database idea interesting,
I would like to hear from you. I need know if such a project would be useful
to other developers. I would hate to spend all the time this project would
take (which would be my own free time) only to find that noone's interested
in using it.

Please post your comments, questions, opinions, observations, concerns, and
constructive flames.

Thanks,
David Hilsabeck
[e-mail]
[e-mail]



+++++++++++++++++++++++++++

>From [e-mail] (AndrewWelc)
Date: 2 Sep 1995 21:10:45 -0400
Organization: America Online, Inc. (1-800-827-6364)

> This should be enough to get the ball rolling. If you understand my
> preferences frustration and find the Preferences Database idea
> interesting, I would like to hear from you. I need know if such a
> project would be useful to other developers. I would hate to spend all
> the time this project would take (which would be my own free time) only
> to find that noone's interested in using it.

Sounds good to me -- let me know if you need any help.

Regards,

Andrew Welch
Thaumaturgist
Ambrosia Software, Inc.

..........

For the latest versions of our software, technical support, and Ambrosia
news, stop by and visit the Ambrosia Software, Inc. support forums:

America Online ---> Keyword: Ambrosia
    CompuServe ---> GO word: Ambrosia
        eWorld --> Shortcut: Ambrosia

+++++++++++++++++++++++++++

>From [e-mail] (Gregory Lo)
Date: Sun, 03 Sep 1995 00:34:48 -0400
Organization: Queen's University

In article <429o59$[e-mail]>, [e-mail]
(HILSABECK DAVID AL) wrote:
> All of the problems preferences pose could easily be overcome if there
> were an automatic method of storing and retrieving data associated with an
> application. Windows programmers have had a solution (a very limited
> solution) to the preferences problem for some time. With .INI files, a
> Windows programmer simply called one of four functions to store and
> retrieve values in what amounted to preferences files. Now with Windows95
> and Windows NT, Windows programmers have the Registration Database, which is
> sort of like preferences on excessive steroids. Even the NeXT operating
> system had a database for storing preferences, even if it did become corrupt
> from time to time.

How about extending Internet Config to handle preferences other than those
associated with Internet applications?
At this point it really doesn't matter how the proposed Preferences
Manager works internally;  it could be ASLM, Component, CFM, staticly
linked libs, etc.
We can get started by determining an interface, which should be
independant of how the manager is implemented.  Internet Config should
definitely be examined for inspiration.

GLo

- ---------------------------------------------------------
Gregory Lo  GLo ?:^(>      <mailto:[e-mail]>

+++++++++++++++++++++++++++

>From [e-mail] (Peter N Lewis)
Date: Sun, 03 Sep 1995 23:37:18 +0800
Organization: Stairways Software

In article <[e-mail]>,
[e-mail] (Gregory Lo) wrote:

>How about extending Internet Config to handle preferences other than those
>associated with Internet applications?

IC already supports anon-internet preferences.  You can store any
preferences you want in there, there is even a defined way to avoid
preference keys from clashing with other developers.

However, IC's actual implementation for storing the prefs is not really
stressed for handling a very large number of preferences, since it just
uses the Resource Manager ("The Resource Manager is not a database" "yes
it is!" "well, not a very good one" "too true!" :-).  However, IC's API
and Component Manager implementation is perfectly well suited to be a
general purpose preference manager.  Now, if someone wanted to take over
and stick a more robust database on the back end of IC... :-)
   Peter.
--
"there is no significant correlation to violence on TV and agression in
daily life" - NHR Broadcasting Culture Research Institute in Tokyo.

---------------------------

>From "Andrew C. Plotkin" <[e-mail]>
Subject: QuickTime volumes
Date: Fri,  1 Sep 1995 13:46:04 -0400
Organization: Carnegie Mellon, Pittsburgh, PA

(This is one of those "it works, but is it a good idea?" questions.)

I have a MIDI file. I convert it to a QT movie using SimpleText 1.3 (I
have QuickTime 2.0 and the QT musical instruments installed, of
course.)

The result is a QT movie with one track, an audio track, and it does
indeed play music. But it's awfully quiet.

So I'd like to amplify it. But both GetMovieVolume (on the movie) and
GetTrackVolume (on the track) return 1.0. That's the value 255, so
actually I guess they're returning 0.998 or something.

The documentation on SetMovieVolume and SetTrackVolume say that they
take fixed-point values from -1.0 to 1.0. Now if I do
    SetMovieVolume(mov, GetMovieVolume(mov) * 2);
it does get louder.  In fact I can crank the multiplier up to about 5;
beyond that it doesn't get any louder, although it still doesn't seem
to produce any errors. But technically I'm doing something undefined
by the documentation.

To confuse matters more, I have a version of MoviePlayer (2.0b2) which
allows me to set the track volume from 0 % to 200 %. I assume that
setting 200% is the equivalent of
    SetTrackVolume(trak, GetTrackVolume(trak) * 2);
So is this sort of legitimized? Is there some documentation on how
high I can crank this thing? Am I running the risk that in some future
(or past) version of QuickTime, it won't get any louder at all?

--Z

"And Aholibamah bare Jeush, and Jaalam, and Korah: these were the borogoves..."

+++++++++++++++++++++++++++

>From [e-mail] (Lawrence D'Oliveiro)
Date: Mon, 04 Sep 1995 16:52:08 +1200
Organization: University of Waikato

In article <[e-mail]>, "Andrew C. Plotkin"
<[e-mail]> wrote:

>So I'd like to amplify it. But both GetMovieVolume (on the movie) and
>GetTrackVolume (on the track) return 1.0. That's the value 255, so
>actually I guess they're returning 0.998 or something.

Actually, I think there may be a slight inconsistency here. I believe the
documentation indicates that the volume is an 8.8 fixed number, so 100%
should really be 256, not 255. But, as you've noticed, the default volume
setting for new movies is 255.

On the other hand, it's been like this since QuickTime 1.0, so they
probably figured it's too late to change now...

>The documentation on SetMovieVolume and SetTrackVolume say that they
>take fixed-point values from -1.0 to 1.0. Now if I do
>    SetMovieVolume(mov, GetMovieVolume(mov) * 2);
>it does get louder.  In fact I can crank the multiplier up to about 5;
>beyond that it doesn't get any louder, although it still doesn't seem
>to produce any errors. But technically I'm doing something undefined
>by the documentation.

In the release notes for QuickTime 1.6, it was made clear that it did now
support volumes greater than 100%, but you needed Sound Manager 3.0 (or
later) in order for this to work.

If you hold down the shift key and click on the volume slider in a movie,
you'll see that the volume range goes up to 300%.

>To confuse matters more, I have a version of MoviePlayer (2.0b2) which
>allows me to set the track volume from 0 % to 200 %. I assume that
>setting 200% is the equivalent of
>    SetTrackVolume(trak, GetTrackVolume(trak) * 2);
>So is this sort of legitimized? Is there some documentation on how
>high I can crank this thing? Am I running the risk that in some future
>(or past) version of QuickTime, it won't get any louder at all?

I'm sure Sound Manager 3.x can apply much larger amplification factors
than this--it's just multiplying numbers, after all. There are really only
two potential problems:

1) If you're trying to amplify a sound that's too soft, it is liable to
get noisy. (This will only apply to sampled-sound tracks, not music
tracks.)
2) If you amplify a sound (or mixture of sounds) that _isn't_ too soft, it
will start to clip and sound horribly distorted.

Apart from watching out for these two things, what you're trying to do IS
in fact fully supported, and should continue to remain so.

PS: if you want to make the volume change permanent, you can use
SetMoviePreferredVolume.

+++++++++++++++++++++++++++

>From [e-mail] (Reinder Verlinde)
Date: Mon, 04 Sep 1995 11:53:57 +0200
Organization: Rijksuniversiteit Utrecht

In article <[e-mail]>, "Andrew C. Plotkin"
<[e-mail]> wrote:

>
> The documentation on SetMovieVolume and SetTrackVolume say that they
> take fixed-point values from -1.0 to 1.0. Now if I do
>     SetMovieVolume(mov, GetMovieVolume(mov) * 2);
> it does get louder.  In fact I can crank the multiplier up to about 5;
> beyond that it doesn't get any louder, although it still doesn't seem
> to produce any errors. But technically I'm doing something undefined
> by the documentation.
>
It is defined in tech notes released with some upgrade. I think that
Sound Manager 3.0 allows 'overdriving' the sound volume, so you should
try to find the release notes for Sound Manager 3.0. I believe the new
maximum sound volume is 0x1FF, but am not sure about it. There also
is a new call which allows you to set volumes for left and right
channels separately (to get cool pseudo-stereo sound movement)

Reinder Verlinde

Unix: the original thing wasn't designed at all.
      Then, when it got designed, it got designed by a committee.
      Worse yet, it got designed by an army of committees.

---------------------------

>From [e-mail] (Eric Klopfer)
Subject: Random Numbers
Date: Fri, 01 Sep 1995 22:18:08 -0500
Organization: U.W. Madison

I need to generate random numbers at a fairly fine scale on the range
0->1.  I am writing my program in Symantec C++ on a PPC.  I have been
using the built in rand() statement, but it only generates 32768 numbers
and this is not a fine enough scale for my use.  I know that the toolbox
has a routine to generate twice this many, but this is not a substantial
enough increase.  I need 10x or 100x better resolution.

Can anyone give me a good (pseudo) random-number generator that won't slow
things down much (I use the random number generator a lot), that would
allow me this increase in resolution?

Thanks.
>From [e-mail] (Eric Klopfer)
Subject: Random Numbers
Date: Fri, 01 Sep 1995 22:18:08 -0500
Organization: U.W. Madison

I need to generate random numbers at a fairly fine scale on the range
0->1.  I am writing my program in Symantec C++ on a PPC.  I have been
using the built in rand() statement, but it only generates 32768 numbers
and this is not a fine enough scale for my use.  I know that the toolbox
has a routine to generate twice this many, but this is not a substantial
enough increase.  I need 10x or 100x better resolution.

Can anyone give me a good (pseudo) random-number generator that won't slow
things down much (I use the random number generator a lot), that would
allow me this increase in resolution?

Thanks.

+++++++++++++++++++++++++++

>From Keith Wiley <[e-mail]>
Date: Sat, 2 Sep 1995 00:59:49 -0400
Organization: University of Maryland College Park

On Fri, 1 Sep 1995, Eric Klopfer wrote:

> I need to generate random numbers at a fairly fine scale on the range
> 0->1.  I am writing my program in Symantec C++ on a PPC.  I have been
> using the built in rand() statement, but it only generates 32768 numbers
> and this is not a fine enough scale for my use.  I know that the toolbox
> has a routine to generate twice this many, but this is not a substantial
> enough increase.  I need 10x or 100x better resolution.
>
> Can anyone give me a good (pseudo) random-number generator that won't slow
> things down much (I use the random number generator a lot), that would
> allow me this increase in resolution?

Assuming you've got memory allocation for really large numbers all
figured out, why don't you generate two numbers and multiply them.  This
will give you any number from 0 - 1073741824.  I'm fairly certain various
numbers in this sequence would be more likely than others, however, as
you approach the more centrally-based numbers, but if that ok, this is a
wonderful solution.  Two calls to rand() won't take up too much more
computation.

. . .. ... ..... ........ ............. .....................
.. ... ..... ....... ........... ............. .................
. .. .... ........ ................ ................................
Keith Wiley, Electrogenetic Engineer             *
University of Maryland at College Park            *   * *     *   *      *
email:  [e-mail]                      ***    **   * *    **     *
world wide web:  http://www.wam.umd.edu/~keithw        *     **   **    ***


+++++++++++++++++++++++++++

>From [e-mail] (David M. Palmer)
Date: 2 Sep 1995 09:54:44 -0400
Organization: Clark Internet Services, Inc., Ellicott City, MD USA

Keith Wiley <[e-mail]> writes:

>On Fri, 1 Sep 1995, Eric Klopfer wrote:

>> I need to generate random numbers at a fairly fine scale on the range
>> 0->1.  I am writing my program in Symantec C++ on a PPC.  I have been

>Assuming you've got memory allocation for really large numbers all
>figured out, why don't you generate two numbers and multiply them.  This
>will give you any number from 0 - 1073741824.  I'm fairly certain various

Bad idea.  In general, you don't get random numbers by doing random things.

Even built-in random number geenerators are often unsuitable for serious
work, especially those that use the linear-congruential method (where
Newrand = (oldrand * A + B) % C ).

Look in 'Numerical Recipies in C' by Teukolsky (sp?) et al. for
good random number generators.

--
                        David Palmer
                        [e-mail]
                        [e-mail]

+++++++++++++++++++++++++++

>From Jonathan Salz <[e-mail]>
Date: 3 Sep 1995 02:50:03 GMT
Organization: Grex

In article <[e-mail]> Eric Klopfer,
[e-mail] writes:
>I need to generate random numbers at a fairly fine scale on the range
>0->1.  I am writing my program in Symantec C++ on a PPC.  I have been
>using the built in rand() statement, but it only generates 32768 numbers
>and this is not a fine enough scale for my use.  I know that the toolbox
>has a routine to generate twice this many, but this is not a substantial
>enough increase.  I need 10x or 100x better resolution.
>
>Can anyone give me a good (pseudo) random-number generator that won't
slow
>things down much (I use the random number generator a lot), that would
>allow me this increase in resolution?

The other respondees to this article are right--rand() is not a terrific
random
number generator. Even if you're satisfied with rand(), multiplying the
results
of two calls to rand() won't give you good random numbers, because
numbers with
lots of factors will pop up often and prime number won't pop up at all.
You
can, however, get random numbers from 0 to 2^30 if you just bitshift the
result
of one call to rand() by 15 to the left, and add the result of another
call to
rand():

unsigned long rand30() { return (rand() << 15) + rand(); }

- Jonathan Salz
- [e-mail]

+++++++++++++++++++++++++++

>From [e-mail] (Patrick L. Reilly)
Date: Sat, 02 Sep 1995 14:51:17 -0700
Organization: IsoQuantic Technologies, LLC

In article <[e-mail]>,
[e-mail] (Eric Klopfer) wrote:

| I need to generate random numbers at a fairly fine scale on the range
| 0->1.  I am writing my program in Symantec C++ on a PPC.  I have been
| using the built in rand() statement, but it only generates 32768 numbers
| and this is not a fine enough scale for my use.  I know that the toolbox
| has a routine to generate twice this many, but this is not a substantial
| enough increase.  I need 10x or 100x better resolution.
|
| Can anyone give me a good (pseudo) random-number generator that won't slow
| things down much (I use the random number generator a lot), that would
| allow me this increase in resolution?
|
| Thanks.


Try this on for size:


#include <stdio.h>
#include <math.h>
#include <console.h>

/*
   Prime modulus multiplicative linear congruential generator
*/
#define     MODULUS  2147483647
#define     MULT1 24112
#define     MULT2 26143

static   long  zrng[] = {0,
1973272912,     281629770,      20006270,       1280689831,
2096730329,     1933576050,
913566091,      246780520,      1363774876,     604901985,
1511192140,     1259851944,
824064364,      150493284,      242708531,      75253171,
1964472944,     1202299975,
233217322,      1911216000,     726370533,      403498145,
993232223,      1103205531,
762430696,      1922803170,     1385516923,     76271663,
413682397,      726466604,
336157058,      1432650381,     1120463904,     595778810,
877722890,      1046574445,
68911991,       2088367019,     748545416,      622401386,
2122378830,     640690903,
1774806513,     2132545692,     2079249579,     78130110,
852776735,      1187867272,
1351423507,     1645973084,     1997049139,     922510944,
2045512870,     898585771,
243649545,      1004818771,     773686062,      403188473,
372279877,      1901633463,
498067494,      2087759558,     493157915,      597104727,
1530940798,     1814496276,
536444882,      1663153658,     855503735,      67784357,
1432404475,     619691088,
119025595,      880802310,      176192644,      1116780070,
277854671,      1366580350,
1142483975,     2026948561,     1053920743,     786262391,
1792203830,     1494667770,
1923011392,     1433700034,     1244184613,     1147297105,
539712780,      1545929719,
190641742,      1645390429,     264907697,      620389253,
1502074852,     927711160,
364849192,      2049576050,     638580085,      547070247};

double   random(stream)
int   stream;
{
   long  zi, lowprd, hi31;

   zi = zrng[stream];
   lowprd = (zi & 65535) * MULT1;
   hi31 = (zi >> 16) * MULT1 + (lowprd >> 16);
   zi = ((lowprd & 65535) - MODULUS) +
         ((hi31 & 32767) << 16) + (hi31 >> 15);
   if(zi < 0)  zi += MODULUS;
   lowprd = (zi & 65535) * MULT2;
   hi31 = (zi >> 16) * MULT2 + (lowprd >> 16);
   zi = ((lowprd & 65535) - MODULUS) +
         ((hi31 & 32767) << 16) + (hi31 >> 15);
   if(zi < 0)  zi += MODULUS;
   zrng[stream] = zi;
   return ((zi >> 7 | 1) + 1)/16777216.0;
}

--
IsoQuantic Technologies   | 1857 W. Calle Del Norte, Chandler, AZ 85224
Patrick L. Reilly,Gen.Mgr.| Voice: +1.602.917.9543 Fax: [phone]
[e-mail]    |   Network Architecture, Design & Analysis

+++++++++++++++++++++++++++

>From Mark Williams <[e-mail]>
Date: Sun, 03 Sep 95 15:46:20 GMT
Organization: Streetly Software


In article <42b54r$[e-mail]>, Jonathan Salz writes:

>
> In article <[e-mail]> Eric Klopfer,
> [e-mail] writes:
> >I need to generate random numbers at a fairly fine scale on the range
> >0->1.  I am writing my program in Symantec C++ on a PPC.  I have been
> >using the built in rand() statement, but it only generates 32768 numbers
> >and this is not a fine enough scale for my use.  I know that the toolbox
> >has a routine to generate twice this many, but this is not a substantial
> >enough increase.  I need 10x or 100x better resolution.
> >
> >Can anyone give me a good (pseudo) random-number generator that won't
> slow
> >things down much (I use the random number generator a lot), that would
> >allow me this increase in resolution?
>
> The other respondees to this article are right--rand() is not a terrific
> random
> number generator. Even if you're satisfied with rand(), multiplying the
> results
> of two calls to rand() won't give you good random numbers, because
> numbers with
> lots of factors will pop up often and prime number won't pop up at all.
> You
> can, however, get random numbers from 0 to 2^30 if you just bitshift the
> result
> of one call to rand() by 15 to the left, and add the result of another
> call to
> rand():
>
> unsigned long rand30() { return (rand() << 15) + rand(); }

I doubt this will work either. I dont know the algorithm rand uses, but I suspect that it will have
a fairly short cycle length (certainly much less than 2^30), and since rand30 will halve this cycle
length, you will get a fairly sparse set of results.

- --------------------------------------
Mark Williams<[e-mail]>

+++++++++++++++++++++++++++

>From [e-mail] (Eric Klopfer)
Date: Mon, 04 Sep 1995 14:42:30 -0500
Organization: U.W. Madison

I got the following response about one of the techniques posted.  I
thought others might be interested as well.

Please refer to:

Simulation Modeling & Analysis, Law/Kelton, McGraw-Hill, 1990, ISBN
0-07-036698-5

for more information about the random number generator I (Patrick L.
Reilly) posted. This book
should be read by anyone planning a simulation activity using random number
generators. Unfortunately, many folks think that just calling the system
rand() function is sufficiently random. Pity the client who relies upon the
results of such a simulation.

Essentially, the number is the length of the random number cycle and the
largest possible number that can be represented on a 32-bit computer. The
list of numbers *are* seeds for the 100 random number streams that are
possible with this code. This way you can have different random numbers
generators for different processes in the same model. The streams are
around 100,000 numbers apart and specially selected to reduce stream
overlap and the few badly correlated cycles in this type of random number
generator.

---------------------------

>From [e-mail] (Mark Coniglio)
Subject: [Q] Gestalt to determine number of serial ports?
Date: Mon, 28 Aug 1995 18:41:48 -0500
Organization: Troika Ranch

I am updating an app that uses the serial driver on the powerbooks. I
would like to be able to determine if there are two serial ports (e.g.,
140, 160, 180, etc.) or just one (e.g., 500 series). In the "Gestalt
Selectors List 2.9" which has appeared in this newsgroup, there is a
'port' selector that seems that it would provide this information. It is
documented there as:

     gestaltPortAttr?           = 'port';
     gestaltHasInternalModem?   = 0?;
     gestaltHasExternalPrinter? = 1?;
     gestaltHasExternalModem?   = 2?;

In fact, this call does not always return a 1 in the third bit if there is
a printer port. Though undocumented, I verified that this 'port' selector
is used by the Chooser in my copy of system 7.5 on my desktop Mac by
setting an A-Trap Break in MacsBug.

Though a recognized Gestalt selector would be the ideal solution, any
other reliable (and machine independent) methods for determining the
number of ports would be useful.

Please email to [e-mail] or post to this newsgroup.

Thanks a million
Mark Coniglio

--
mail: [e-mail]
http: www.art.net/Studios/Performance/Dance/Troika_Ranch/TroikaHome.html

+++++++++++++++++++++++++++

>From [e-mail] (Christophe ANDRES)
Date: Tue, 29 Aug 1995 22:47:17 +0200
Organization: Chrisoft

In article <[e-mail]>,
[e-mail] (Mark Coniglio) wrote:

>
> Though a recognized Gestalt selector would be the ideal solution, any
> other reliable (and machine independent) methods for determining the
> number of ports would be useful.
>
> Please email to [e-mail] or post to this newsgroup.
>

I don't know of a gestalt selector counting the serial ports, but using
the comm Toolbox, in one of my programs, I have to get them to show a
list. I think this could be of help.

P.S. : this is an taken directly from a MacApp programm.

{
   CRMRec      c;
   CRMRecPtr cPtr = &c;
   CRMSerialPtr serialPtr;
   comPortID theComPort;

   c.crmDeviceType = crmSerialDevice;  // to look only for serial ports
   c.crmDeviceID = 0;
   while (cPtr != nil)  // Browse through all entries
      {
         cPtr = CRMSearch(cPtr);
         if (cPtr)
            {
               serialPtr = (CRMSerialPtr)cPtr->crmAttributes;
               c.crmDeviceID = cPtr->crmDeviceID;

               theComPort.portName = CStr255(*(serialPtr->name));
               theComPort.portIcon = serialPtr->deviceIcon;
               this->InsertElementsBefore(fSize + 1, &theComPort, 1);
            }
      }
}

--
Chrisoft
>From [e-mail] (Mark Coniglio)

+++++++++++++++++++++++++++

>From [e-mail] (Jerome Jahnke)
Date: Mon, 4 Sep 1995 17:14:53 GMT
Organization: BSD Academic Computing

In article <[e-mail]>,
[e-mail] (Mark Coniglio) wrote:

> I am updating an app that uses the serial driver on the powerbooks. I
> would like to be able to determine if there are two serial ports (e.g.,
> 140, 160, 180, etc.) or just one (e.g., 500 series). In the "Gestalt
> Selectors List 2.9" which has appeared in this newsgroup, there is a
> 'port' selector that seems that it would provide this information. It is
> documented there as:
>
>      gestaltPortAttr?           = 'port';
>      gestaltHasInternalModem?   = 0?;
>      gestaltHasExternalPrinter? = 1?;
>      gestaltHasExternalModem?   = 2?;
>
> In fact, this call does not always return a 1 in the third bit if there is
> a printer port. Though undocumented, I verified that this 'port' selector
> is used by the Chooser in my copy of system 7.5 on my desktop Mac by
> setting an A-Trap Break in MacsBug.

Use the comm toolbox. It has the CRMSerialRecords that are a list of all
serial connections your machine can make. You can then sort through them.
It makes your software much more portable. It also includes snazzy icon
families and real text descriptions of the names of the ports.

Jer,

--
Jerome Jahnke
BSD Academic Computing
[e-mail]

---------------------------

End of C.S.M.P. Digest
**********************