Developer CD Series 1994…tember: Reference Library/Dev.CD Sep 94.toast/Technical Documentation/C.S.M.P. Digests/csmp-digest-v3-034/doubleCR.1

Text File | 1994-07-07 | 66.3 KB | 1,787 lines

C.S.M.P. Digest             Sun, 12 Jun 94       Volume 3 : Issue 34

Today's Topics:

        32-Bit Addressing
        AppleScript Questions
        Code Warrior allocation of Objects
        Code Warrior: ruminations of a long-time MPW user
        Refinement of SC++ 7.0 code gen bug
        Serial Port Programming?
        Serial port use
        THINK Class Library FTP Site?
        storage for XCMDS



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

The digest is a collection of article threads from the internet newsgroup
comp.sys.mac.programmer.  It is designed for people who read c.s.m.p. 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]. Currently no previous volumes of the CSMP
digest are available there.

Also, the digests are available to WAIS users.  To search back issues
with WAIS, use comp.sys.mac.programmer.src. With Mosaic, use
http://www.wais.com/wais-dbs/comp.sys.mac.programmer.html.


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

>From [e-mail] (Spencer Low)
Subject: 32-Bit Addressing
Date: 20 May 1994 18:43:48 -0700
Organization: LowTek Creations

How do you detect whether 32-bit Addressing is enabled/disabled? Also,
how can you turn it on? I remember reading somewhere that there's a bit
in PRAM that you can modify. Does anyone have any specifics or sample
code (THINK C)? I realize that the Mac would be require to be rebooted,
but I would still like an automated way to turn it on (with the
authorization of the user).

Thanks,
Spencer


--
  Spencer Low       | Creators of MaxRAM!!
  LowTek Creations  | Anonymous FTP to crl.com and look in
  [e-mail]  | /users/ro/spencerl/ for more info.

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

>From [e-mail] (John W. Baxter)
Date: Fri, 20 May 1994 23:12:09 -0700
Organization: Internet for the Olympic Peninsula

In article <2rjp0k$[e-mail]>, [e-mail] (Spencer Low) wrote:

> How do you detect whether 32-bit Addressing is enabled/disabled? Also,
> how can you turn it on? I remember reading somewhere that there's a bit
> in PRAM that you can modify. Does anyone have any specifics or sample
> code (THINK C)? I realize that the Mac would be require to be rebooted,
> but I would still like an automated way to turn it on (with the
> authorization of the user).

Send a 4 byte -1 through the StripAddress trap, and see what comes out.
--
John Baxter    Port Ludlow, WA, USA  [West shore, Puget Sound]
   [e-mail]

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

>From [e-mail] (Spencer Low)
Date: 21 May 1994 00:08:14 -0700
Organization: LowTek Creations

John W. Baxter ([e-mail]) wrote:
> Send a 4 byte -1 through the StripAddress trap, and see what comes out.

I just tried that and StripAddress returned 0xFFFFFFFF, or -1. What will
it return in 24-bit mode? 0x00FFFFFF?

I think I just answered my own question (well, part of it)...
GestaltEqu.h tells me that the gestaltAddressingModeAttr selector will
tell me about the addressing mode of the Mac. Just test for bit 0
(gestalt32BitAddressing).

So how do I turn it on? And I shouldn't turn it on if gestalt32BitCapable
is false?

Thanks,
Spencer
--
  Spencer Low       | Creators of MaxRAM!!
  LowTek Creations  | Anonymous FTP to crl.com and look in
  [e-mail]  | /users/ro/spencerl/ for more info.

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

>From [e-mail] (Spencer Low)
Date: 21 May 1994 12:47:41 -0700
Organization: LowTek Creations

Spencer Low ([e-mail], aka "me") wrote:
> So how do I turn it on? And I shouldn't turn it on if gestalt32BitCapable
> is false?

I did soom searching on the 'net (the alt.sources.mac archive) and I
found out some information about extended PRAM (XPRAM) and memory
settings. I looked at the Extended PRAM Layout from Kiselyov Oleg
([e-mail]) and I copied/composed the following code:

(THINK C style with inline assembly)

char    buffer;

asm {
    LEA    buffer, A0    // pass the address as a parameter
    MOVE.W    #1, D0        // the length of the data
    SWAP    D0
    MOVE.W    #0x8A, D0    // location in XPRAM
    DC.W    0xA051        // ReadXPRam trap
}
BitSet( &buffer, 15 );    // Set xxxx x1x1 (should mean 32-bit on)
BitSet( &buffer, 13 );
asm {
    LEA    buffer, A0    // pass the address as a parameter
    MOVE.W    #1, D0        // the length of the data
    SWAP    D0
    MOVE.W    #0x8A, D0    // location in XPRAM
    DC.W    0xA052        // WriteXPRam trap
}

Kiselyov Oleg's info tells us the following:

Loc  Length
(hex)  (dec)
8A      1               Bit field: Memory/cache control flags
           1xxx xxxx    68040 Cache is OFF | Toggled through Cache control
           0xxx xxxx    68040 Cache is ON  | panel
           xxxx x1x1    32-bit addressing is ON (toggled through the Memory
                        control panel)

So it seems like my code is setting the right stuff, correct? Does my
code make sense? I can't test it on my 660AV, since the 660AV doesn't
support 24-bit mode (or does it?). Also, I noticed that Kiselyov's info
says 1xxx xxxx means that the cache is off, but on my Mac, my cache is on
and ReadXPRam returns 1xxx xxxx. Is there a typo or something?

Thanks in advance for your help,

Spencer
--
  Spencer Low       | Creators of MaxRAM!!
  LowTek Creations  | Anonymous FTP to crl.com and look in
  [e-mail]  | /users/ro/spencerl/ for more info.

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

>From [e-mail] (Spencer Low)
Date: 21 May 1994 14:00:02 -0700
Organization: LowTek Creations

Spencer Low ([e-mail]) wrote:
> BitSet( &buffer, 15 );    // Set xxxx x1x1 (should mean 32-bit on)
> BitSet( &buffer, 13 );

Whoops! This code should be BitSet(&buffer, 5); & BitSet(&buffer, 7);
instead. (this is the first time I've ever used BitSet.)

> Kiselyov Oleg's info tells us the following:

> Loc  Length
> (hex)  (dec)
> 8A      1               Bit field: Memory/cache control flags
>            1xxx xxxx    68040 Cache is OFF | Toggled through Cache control
>            0xxx xxxx    68040 Cache is ON  | panel
>            xxxx x1x1    32-bit addressing is ON (toggled through the Memory
>                         control panel)

> So it seems like my code is setting the right stuff, correct? Does my
> code make sense? I can't test it on my 660AV, since the 660AV doesn't
> support 24-bit mode (or does it?). Also, I noticed that Kiselyov's info
> says 1xxx xxxx means that the cache is off, but on my Mac, my cache is on
> and ReadXPRam returns 1xxx xxxx. Is there a typo or something?

It seems like I was the one making the typo. I checked on my calculator
(conversions) and ReadXPRam was returning x1xx x1x1, meaning that 32-bit
addressing was on and the cache was on. I don't know what that second "1"
means, though. Perhaps it means that 32-bit can't be turned off?

Spencer
--
  Spencer Low       | Creators of MaxRAM!!
  LowTek Creations  | Anonymous FTP to crl.com and look in
  [e-mail]  | /users/ro/spencerl/ for more info.

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

>From [e-mail] (Bob Boonstra)
Date: Sun, 22 May 1994 07:37:29 -0500
Organization: MITRE

In article <2rjp0k$[e-mail]>, [e-mail] (Spencer Low) wrote:

> How do you detect whether 32-bit Addressing is enabled/disabled? Also,
> how can you turn it on?

mode = GetMMUMode();  /* returns false32b (0) or true32b (1) */
SwapMMUMode(&mode);  /* sets addressing mode and returns old mode */
--
-- Bob Boonstra
-- [e-mail]
-- My opinion, not my employer's

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

>From [e-mail] (Spencer Low)
Date: 22 May 1994 12:56:20 -0700
Organization: LowTek Creations

Bob Boonstra ([e-mail]) wrote:
> In article <2rjp0k$[e-mail]>, [e-mail] (Spencer Low) wrote:

> > How do you detect whether 32-bit Addressing is enabled/disabled? Also,
> > how can you turn it on?

> mode = GetMMUMode();  /* returns false32b (0) or true32b (1) */
> SwapMMUMode(&mode);  /* sets addressing mode and returns old mode */

Is this the same 32-bit mode that the Memory control panel controls? I
always thought that this was a different related thing. If there are
these routines, why do we have to reboot after changing the mode in the
Memory control panel?

Thanks,
Spencer
--
  Spencer Low       | Creators of MaxRAM!!
  LowTek Creations  | Anonymous FTP to crl.com and look in
  [e-mail]  | /users/ro/spencerl/ for more info.

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

>From [e-mail] (John W. Baxter)
Date: Mon, 23 May 1994 07:47:10 -0700
Organization: Internet for the Olympic Peninsula

In article <2rodd4$[e-mail]>, [e-mail] (Spencer Low) wrote:

> Bob Boonstra ([e-mail]) wrote:
> > In article <2rjp0k$[e-mail]>, [e-mail] (Spencer Low) wrote:
>
> > > How do you detect whether 32-bit Addressing is enabled/disabled? Also,
> > > how can you turn it on?
>
> > mode = GetMMUMode();  /* returns false32b (0) or true32b (1) */
> > SwapMMUMode(&mode);  /* sets addressing mode and returns old mode */
>
> Is this the same 32-bit mode that the Memory control panel controls? I
> always thought that this was a different related thing. If there are
> these routines, why do we have to reboot after changing the mode in the
> Memory control panel?

SwapMMUMode () is intended for *temporary* mode changes (such as the
necessary switch to 32-bit mode to access video memory on typical cards).
It does not magically change all the heaps in the machine between 24-bit
form and 32-bit form (not to mention that there are now [at least] two
variations of the 32-bit heap).  [So...what you can do during the temporary
switch is quite limited.]

--
John Baxter    Port Ludlow, WA, USA  [West shore, Puget Sound]
   [e-mail]

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

>From [e-mail] (Philip Machanick)
Date: Wed, 25 May 1994 07:32:57 +0200
Organization: Computer Science Dept, U of Witwatersrand

In article <2rodd4$[e-mail]>, [e-mail] (Spencer Low) wrote:

> Is this the same 32-bit mode that the Memory control panel controls? I
> always thought that this was a different related thing. If there are
> these routines, why do we have to reboot after changing the mode in the
> Memory control panel?

Could it be that there's a difference between locally changing the mode for
a little piece of your code (e.g., check if in 32-bit mode, if so switch it
off for a piece of old code then switch it on again) and globally changing
it for every piece of software on the system?
--
Philip Machanick                   [e-mail]
Department of Computer Science, University of the Witwatersrand
2050 Wits, South Africa
phone [phone]  fax [phone]

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

>From [e-mail] (Dieter Shirley)
Date: Fri, 27 May 1994 17:27:04 GMT
Organization: Cognos Incorporated, Ottawa CANADA

Spencer Low ([e-mail]) wrote:
: Bob Boonstra ([e-mail]) wrote:
: > mode = GetMMUMode();  /* returns false32b (0) or true32b (1) */
: > SwapMMUMode(&mode);  /* sets addressing mode and returns old mode */

: Is this the same 32-bit mode that the Memory control panel controls? I
: always thought that this was a different related thing. If there are
: these routines, why do we have to reboot after changing the mode in the
: Memory control panel?

All the traps and stuff are initialized at boot up to either work in
24-bit or 32-bit mode.  If you swap it, you can do writes to upper
(video and NuBus) memory, but if you call any traps during this time,
they'll expect to be in the mode that the computer started in.  If you're
in a different mode, they could crash horrifically.

If you need to know if 32-bit addressing is on, use the Gestalt call.
If you need 32-bit addressing on for a *short* time use SwapMMUMode().
If you need 32-bit addressing on for a long time (like, the whole
course of your program) inform the user to turn it on via the control
panel, and then quit.  You could set the bit in PRAM yourself, but
then they'd still have to restart.

Good luck,
-dete

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

>From Stephan Bublava <[e-mail]>
Subject: AppleScript Questions
Date: 14 May 1994 17:45:37 GMT
Organization: Vienna University of Technology

Hi, I have two questions regarding AppleScript:

1) Is there any way to allow/disallow user interaction?

2) Can I run a script asynchronously? I am playing around
   with OSA menu and whenever I activate a script the
   front application is blocked until the script is finished.
   I know that "ignoring application responses" works around
   this, but on the other hand, I need application responses
   for some scripts to work, and to get errors back.

Any info would be appreciated.

Stephan

--
Stephan Bublava
[e-mail]

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

>From Jens Alfke <[e-mail]>
Date: Wed, 25 May 1994 18:38:34 GMT
Organization: Apple Computer

Stephan Bublava, [e-mail] writes:
> 1) Is there any way to allow/disallow user interaction?

I seem to remember that there is, but I can't remember the syntax. Check the
manual...

> 2) Can I run a script asynchronously? I am playing around
>    with OSA menu and whenever I activate a script the
>    front application is blocked until the script is finished.
>    I know that "ignoring application responses" works around
>    this, but on the other hand, I need application responses
>    for some scripts to work, and to get errors back.

If you save a script as an application ("applet") it will run asynchronously
in its own process. If you still want the ease-of-triggering of a menu
command, you can either put the applet in the Apple menu (which will bring
its layer to the front) or write a little launcher script for OSAMenu that
just launches your applet.

--Jens Alfke
  jens_alfke@powertalk              Rebel girl, rebel girl,
            .apple.com              Rebel girl you are the queen of my world

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

>From Stephan Bublava <[e-mail]>
Date: 27 May 1994 12:17:14 GMT
Organization: Vienna University of Technology

In article <[e-mail]> Jens Alfke,
[e-mail] writes:

>> 1) Is there any way to allow/disallow user interaction?

>I seem to remember that there is, but I can't remember the
>syntax. Check the manual...

Well, that's what I did before posting, but it does not contain
anything (I checked the manual for 1.0, and the release notes
for both 1.0  and 1.1).

Thanks for your information on my other question.

In the meantime I have yet another problem: I can't get
transactions to work.

For example:

tell application "FileMaker Pro"
    with transaction
    ...
    end transaction
end tell

The problem is "end transaction" because there also is
a command with the same name. If you only write "end" then
it will compile, but only once because AppleScript changes
the line back to "end transaction".

But apart from that, it didn't work right (with FileMaker Pro):
In the transaction I created a new record and then searched
for a record that did not exist. I expected that the record I
had created would be automatically deleted again (since the
search failed), but I just got an error message.

Is this one of the many problems FileMaker has with AppleScript,
or am I doing something wrong?

Stephan

--
Stephan Bublava
[e-mail]

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

>From [e-mail] (Brian V. Hughes)
Date: 27 May 1994 15:52:27 GMT
Organization: Dartmouth College, Hanover, NH, USA

Stephan Bublava <[e-mail]> writes:

>In the meantime I have yet another problem: I can't get
>transactions to work.

>For example:

>tell application "FileMaker Pro"
>    with transaction
>    ...
>    end transaction
>end tell

>The problem is "end transaction" because there also is
>a command with the same name. If you only write "end" then
>it will compile, but only once because AppleScript changes
>the line back to "end transaction".

    I think what you need to do is use the "Begin Transation" and "End
Transaction" AppleScript commands that are part of FM Pro's
dictionary. I don't know how they differ from the with transaction
control statement but my guess is that AppleEvents were added to FM Pro
before the with transaction control statement was fully implimented, and
therefore it doesn't support it correctly. That's why there are separate
commands for it that are FM Pro specific.

-Hades

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

>From [e-mail] (Jon Pugh)
Date: Sun, 29 May 1994 06:36:13 GMT
Organization: NETCOM On-line Communication Services (408 261-4700 guest)

Jens Alfke ([e-mail]) wrote:
> Stephan Bublava, [e-mail] writes:
> > 1) Is there any way to allow/disallow user interaction?

> I seem to remember that there is, but I can't remember the syntax. Check the
> manual...

There are no user interaction events or settings in AppleScript.  That's
why programs will need to implement something similar to what I did in
PhotoFlash.  I created an application property called "user iteraction"
which takes a set of enums; none, interact with local, interact with remote,
& interact with all.  Then before you put up any dialog, call a CanDoUI
routine which checks the UI level and calls AEInteractWithUser if it is not
none (since the AEM doesn't include a none setting).  This ensures that you
get brought to the front and/or a notification is posted.  You need to
implement none because the AEM's minimum is "interact with local" and if you
run a script (via a script menu or so) your script is local and allowed to
interact.  Yuck.  You need to use a global to get the never interact level
and you need to provide defaults for every function that calls CanDoUI and
gets back a false.

I would suggest that a future version of AppleScript and the AEM support
this feature, if there anyone working on either.  ;)

Jon


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

>From [e-mail] (John W. Baxter)
Date: Sun, 29 May 1994 08:25:58 -0700
Organization: Internet for the Olympic Peninsula

In article <[e-mail]>, [e-mail] (Jon Pugh)
wrote:

> There are no user interaction events or settings in AppleScript.  That's
> why programs will need to implement something similar to what I did in
> PhotoFlash.  I created an application property called "user iteraction"
> which takes a set of enums; none, interact with local, interact with remote,
> & interact with all.  Then before you put up any dialog, call a CanDoUI
> routine which checks the UI level and calls AEInteractWithUser if it is not
> none (since the AEM doesn't include a none setting).  This ensures that you
> get brought to the front and/or a notification is posted.  You need to
> implement none because the AEM's minimum is "interact with local" and if you
> run a script (via a script menu or so) your script is local and allowed to
> interact.  Yuck.  You need to use a global to get the never interact level
> and you need to provide defaults for every function that calls CanDoUI and
> gets back a false.

Jon...The AE Manager is quite happy with "never interact" as sent by the
sender.  It is of course true that the client side's minimum level for
AESetInteractionAllowed () is self (which I think I consider "lower" than
local, since it cuts out other clients).  So you are right that to achieve
never, the application has to remember that it shouldn't interact.

> I would suggest that a future version of AppleScript and the AEM support
> this feature, if there anyone working on either.  ;)

I found it surprising that AppleScript elected to use only "can" interact.
Frontier, on the other hand, can currently be set to send either "never" or
"always" (version 2 was "never" or "can"...the change to "always" has
caused some scripting glitches with FileMaker, which now must be brought to
the front before certain actions are sent from Frontier).

So an OSA Menu script which needs to be never interact could be written in
UserTalk, with the level being set to never [the UserTalk for that is
setEventInteraction (false)...the setting is per-script].

I concur that AppleScript should be enhanced to permit per-script settings
of "never", "can", and "always" (and control might be desired for the
kAECanSwitchLayer flag, too).

Good job with your PhotoFlash workaround...I'll have to remember to
implement (ie, steal) that idea.
--
John Baxter    Port Ludlow, WA, USA  [West shore, Puget Sound]
   [e-mail]

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

>From [e-mail] (Eric Shieh)
Subject: Code Warrior allocation of Objects
Date: 22 May 1994 20:54:17 GMT
Organization: University of California, Berkeley

Just received CW Gold DR/2 2 weeks ago and have ported my program
from Think C 5.0 to it...seems to work great except that the CW
compiler is a lot pickier about types...anyways, that's OK.

The bulk of my program is in standard C.  Now, I'm trying to add some
stuff using C++. The problem: once I allocate an object via 'new' (i haven't
tryied allocating it on the stack), about 80K is allocated in the heap...
Now, this may be OK in a 2-3 meg program, but since my program's total
file size is 68K in a 120K partition, 80K is a big deal...

Oh, and if I call delete, the 80K remains in the heap, non-relocatable. but
calling new again doesn't try to allocate another 80K....

So, my question is: Is this just the way it works or is this a bug? It seems
rather odd that it's allocating so much memory since as I said, my
program size is 68K, what could it be sticking in that 80K block?

Oh, btw, is their any way to create objects as Handles instead of Pointers?

And I also recall a codewarrior mailing list around here, anyone know the
address/instructions?

Thanks for listening,
Eric

[e-mail]


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

>From [e-mail] (Eric Scouten)
Date: Mon, 23 May 1994 13:31:56 GMT
Organization: University of Minnesota, Student Affairs

In article <2rogpp$[e-mail]>, [e-mail]
(Eric Shieh) wrote:

> The bulk of my program is in standard C.  Now, I'm trying to add some
> stuff using C++. The problem: once I allocate an object via 'new' (i haven't
> tryied allocating it on the stack), about 80K is allocated in the heap...

CodeWarrior's operator new bypasses the Memory Manager when allocating
small objects. It allocates its own pool of (apparently) 80K for any
objects that are below a certain threshold size (4K I think?). This pool
remains permanently allocated until the app quits; if it fills, CW grabs
another 80K pool to create the next set of objects.

The advantage of this is that the operator new/delete pair can manage
objects more efficiently (both in memory and speed) than can the Memory
Manager.


> Oh, btw, is their any way to create objects as Handles instead of Pointers?

Not in any compiler that supports multiple inheritance. (At least SC++ 7.0
and CodeWarrior don't support handle objects.) Handles are an ugly way to
manage objects anyway: accessing member fields is considerably less
efficient, and you have to be much more careful about how you use pointers.


> And I also recall a codewarrior mailing list around here, anyone know the
> address/instructions?

Yup. To subscribe send a one line message...

   subscribe cwarrior <Your name in real life, not e-mail address>

... to the address [e-mail]. Make sure your signature is turned
off.

-Eric


--
Eric Scouten  *  Univ of Minnesota  *  Student Affairs  *  [phone]
  ** MS Computer Science student, Univ of Illinois, starting Aug '94 **

WYSIWIG: What You See Is What You Get

WYSIWOBBLE: What You See Isn't What's Output By the Blasted Laserwriter Eh?

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

>From [e-mail] (Vincent Partington)
Date: 24 May 1994 11:58:59 GMT
Organization: FWI, University of Amsterdam

[e-mail] (Eric Shieh) writes:

>The bulk of my program is in standard C.  Now, I'm trying to add some
>stuff using C++. The problem: once I allocate an object via 'new' (i haven't
>tryied allocating it on the stack), about 80K is allocated in the heap...
>Now, this may be OK in a 2-3 meg program, but since my program's total
>file size is 68K in a 120K partition, 80K is a big deal...

This is because new and delete use their own allocator system that gets
chunks of OS-memory at once to efficiently manage memory.
You can bypass this be recompiling the C++ Runtime library: the source-
code for new/delete is in the C++ Runtime directory, there's a define that
you can change to make new/delete use NewPtr/DisposePtr, just have
a look at the code in the New.cp and NewMore.cp files.

Vincent.
--
My opinions are not my own. I copy them    | Internet : [e-mail]
from books, television, video, the net,    |            [e-mail]
my friends, my parents, my teachers and    | FidoNet  : 2:281/202.15
and numerous other contributors.           | NeST     : 90:500/202.15

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

>From Jens Alfke <[e-mail]>
Date: Wed, 25 May 1994 19:15:42 GMT
Organization: Apple Computer

In article <2rsq63$[e-mail]> Vincent Partington,
[e-mail] writes:
> This is because new and delete use their own allocator system that gets
> chunks of OS-memory at once to efficiently manage memory.
> You can bypass this be recompiling the C++ Runtime library...

Naw, just add an 'operator new' and 'operator delete' in the base class and
have it use some other allocation scheme. NewPtr will work okay but will be
lots slower if tons of objects are being created and deleted. If the program
is already using some more efficient memory allocator (malloc or something)
then you can use that.

--Jens Alfke
  jens_alfke@powertalk              Rebel girl, rebel girl,
            .apple.com              Rebel girl you are the queen of my world

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

>From [e-mail] (Vincent Partington)
Date: 26 May 1994 10:47:10 GMT
Organization: FWI, University of Amsterdam

Jens Alfke <[e-mail]> writes:

>> This is because new and delete use their own allocator system that gets
>> chunks of OS-memory at once to efficiently manage memory.
>> You can bypass this be recompiling the C++ Runtime library...

>Naw, just add an 'operator new' and 'operator delete' in the base class and
>have it use some other allocation scheme. NewPtr will work okay but will be
>lots slower if tons of objects are being created and deleted. If the program
>is already using some more efficient memory allocator (malloc or something)
>then you can use that.

When adding those operators to "the" base class you have to change your C++
code and derive all your classes from that class. Changing the C++ Runtime lib
is admittedly dirtier but doesn't require changing the C++ code.
Changing the C++ Runtime lib also allows you to change the allocation routines
for arrays becuase AFAIK you can't change operator new/delete for arrays.

Vincent\n
--
My opinions are not my own. I copy them    | Internet : [e-mail]
from books, television, video, the net,    |            [e-mail]
my friends, my parents, my teachers and    | FidoNet  : 2:281/202.15
and numerous other contributors.           | NeST     : 90:500/202.15

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

>From Jens Alfke <[e-mail]>
Date: Fri, 27 May 1994 20:04:20 GMT
Organization: Apple Computer

Vincent Partington, [e-mail] writes:
> When adding those operators to "the" base class you have to change your C++
> code and derive all your classes from that class. Changing the C++ Runtime
lib
> is admittedly dirtier but doesn't require changing the C++ code.

That's a fair point; but you can always just define your own global operator
new and operator delete, and then include the C++ runtime library without its
operator new support. (I think there is a version of the library without
operator new in it.) All of your classes will use the global operator
new/delete by default.

I'm always leery of modifying system libraries, since they generally change
in every release and then I have to redo my work all over again.

--Jens Alfke
  jens_alfke@powertalk              Rebel girl, rebel girl,
            .apple.com              Rebel girl you are the queen of my world

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

>From [e-mail] (Amanda Walker)
Subject: Code Warrior: ruminations of a long-time MPW user
Date: Tue, 24 May 1994 11:47:08 -0500
Organization: InterCon Systems Corporation, Herndon, VA USA

Several people have expressed surprise at my couple of posts supporting Code
Warrior yesterday, since I've been a die-hard MPW user for years.  I thought
I'd go into some more detail on this, since the issues involved are not unique
to me alone.

I've been using MPW since version 2.  I've also used just about every other C
compiler to come out for the Mac, from the Megamax C and Consulair C to THINK
C 7.0.  All of these systems have their good points and their bad points.
However, for multi-project multi-person development, MPW has held the lead.
It is no speed demon in terms of lines per minute compiled, but it is
extremely flexible.  The abilities to script the environment, provide multi-
user source code and revision control, and add custom tools serve to make up
in the long term for a lack of compilation speed.  Also, Apple generally
releases new technology via MPW header files and/or libraries, which can be
especially important when dealing with pre-release versions of Apple software
(as I have been a number of times).  All of these things have kept me in the
MPW camp for the kind of work I do most.  Symantec C 6.0 looked promising,
since it was finally scriptable, but the bugs in it and the general manner in
which Symantec has handled 6.x and 7.0 have undermined my faith in their
ability to deliver a quality product, despite having used it off and on since
it was called "Lightspeed C".

I'd seen a preview of Code Warrior in December at one of Apple's "PowerPC
Coding Kitchens" which they ran for developers in the seeding program.  It
looked promising, but it was still kind of buggy, and was not scriptable.  The
person doing the demo characterized it as analogous to THINK C 5.x for the
PowerPC.  As a result, I sighed and went back to the Apple RISC SDK, which I
fought with for some time before finally getting a build that I felt
comfortable saying we could ship to customers.  However, by this time I was
getting aggravated at MPW as well.  The PowerPC C compilers and linker in
Apple's SDK are excruciatingly slow, and the optimizer breaks.  TCP/Connect II
1.2.1 is built with optimization off to avoid compiler bus errors and bad code
generation.  It also doesn't run native, making the 8100/80 on my desk almost
usefuless for anything but testing.

It has not been a fun spring in some respects.

By the WWDC, Code Warrior had started to get rave reviews, both here and from
other developers whose opinions I have learned to respect, so I went with an
open mind.  After watching the Metrowerks presentation, and sitting down with
Code Warrior myself, I became convinced that they really did know what they
were doing.  It was fast, native, and had an optimizer that seems to actually
work.  It's scriptable, so that I can do a large multi-project build without
having to hand-drive the process.  It coexists nicely with SourceServer, so
that I can continue to use existing source code databases.  It's code
generator isn't a mangled version of an 8x86 code generator, so it actually
generates nice code.  I don't have to hook my PowerBook up to my Power Mac
just to run the debugger.  It's not perfect, but it finally tipped the scale
compared to MPW.  MPW tool versions of the compilers and linkers, which
Metrowerks says are in the pipeline, will be the final clincher, since they
will give me the best of both worlds.

Besides, Metrowerks seems to be as much a bunch of crazy workaholics as
InterCon is :).


Amanda Walker
Advanced Projects
InterCon Systems Corporation



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

>From [e-mail] (Michael W. Wellman)
Date: Tue, 24 May 1994 14:30:57 -0500
Organization: InterCon Systems Corp., Herndon, VA, USA

[e-mail] (Amanda Walker) writes:
> However, by this time I was getting aggravated at MPW as well.
> The PowerPC C compilers and linker in Apple's SDK are excruciatingly
> slow, and the optimizer breaks.  TCP/Connect II 1.2.1 is built with
> optimization off to avoid compiler bus errors and bad code generation.
> It also doesn't run native, making the 8100/80 on my desk almost
> usefuless for anything but testing.

Er, I'd just like to make it clear that the "It" above refers to MPW and not
TCP/Connect II 1.2.1.  The latter works quite nicely natively...

Now if only Metrowerks and Apple's Open Transport would work out their
differences regards Apple's Shared Library Manager.

mikel





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

>From [e-mail] (Michael Peirce)
Date: Tue, 24 May 94 14:30:05 PST
Organization: Peirce Software, Inc.


In article <[e-mail]> (comp.sys.mac.programmer), [e-mail] (Michael W. Wellman) writes:
> [e-mail] (Amanda Walker) writes:
> > However, by this time I was getting aggravated at MPW as well.
> > The PowerPC C compilers and linker in Apple's SDK are excruciatingly
> > slow, and the optimizer breaks.  TCP/Connect II 1.2.1 is built with
> > optimization off to avoid compiler bus errors and bad code generation.
> > It also doesn't run native, making the 8100/80 on my desk almost
> > usefuless for anything but testing.
>
> Er, I'd just like to make it clear that the "It" above refers to MPW and not
> TCP/Connect II 1.2.1.  The latter works quite nicely natively...

The "prerelease MPW" version 3.4a2 from E.T.O. 14 has some native
parts in it now.

The shell is native, as is make, rez, derez, and rezdet.  Too bad
Link and C aren't native, but the native shell (and my faster hard
disk) make for a quicker startup.  And the Make doesn't take any time
anymore either.

It's only a start, but at least its better than nothing.

(I'm very much looking forward to trying the CodeWarrior compiler
and linker under MPW!)

__ Michael Peirce        __ [e-mail]
__ Peirce Software, Inc. __ 719 Hibiscus Place, Suite 301
__                       __ San Jose, California USA 95117-1844
__ Makers of: Smoothie & __ voice: +1.408.244.6554 fax: [phone]
__    Peirce Print Tools __ AppleLink: peirce & AOL: AFC Peirce

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

>From [e-mail] (Christopher Songer)
Date: Wed, 25 May 1994 13:50:40 GMT
Organization: Lexmark International, Lexington, KY

In article <[e-mail]> [e-mail] (Amanda Walker) writes:

   [...]
   By the WWDC, Code Warrior had started to get rave reviews, both here and from
   other developers whose opinions I have learned to respect, so I went with an
   open mind.
   [...]

Hi!

     Thanks for the review -- it is much appreciated and was quite
informative. Can you speak to the source control management provided
by the CodeWarrior package?

Thanks!
-Chris
[e-mail]

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

>From [e-mail] (Amanda Walker)
Date: Wed, 25 May 1994 17:35:41 -0400
Organization: InterCon Systems Corporation, Herndon, VA USA

[e-mail] (Michael W. Wellman) writes:
> Er, I'd just like to make it clear that the "It" above refers to MPW
> and not TCP/Connect II 1.2.1.  The latter works quite nicely natively...

Yes, indeed.  I used it to post that article, in fact :).


Amanda Walker
InterCon Systems Corporation



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

>From [e-mail] (John Purlia)
Date: Thu, 26 May 1994 13:17:22 -0800
Organization: Qualcomm, Inc.

In article <[e-mail]> [e-mail]
(Amanda Walker) writes:
>[...]
>By the WWDC, Code Warrior had started to get rave reviews, both here and from
>other developers whose opinions I have learned to respect, so I went with an
>open mind.
>[...]

Yes, a very large number of Apple engineers are using and raving about Code
Warrior which is encouraging when evaluating the stability and usefulness
of a new development environment.  I'd very much prefer to use the power
available to me through the MPW environment so I'm eagerly awaiting the
release of native MPW tools from MetroWerks -- especially after talking to
a couple of people at the MPW station at the conference expo who
regrettably informed me that there are no plans for porting a 68K compiler
to run native on PPC, which means... writing fat binaries will require you
to compile the 68K portion of your app with an emulated compiler!  AARGH!!!
...........................................................................
John Purlia          : My brain; not my company's brain.  My brain says...
[e-mail] :
Applelink    AM0470  :   "Newton could never live on Via de la Valle,
NewtonMail:  JPurlia :  but it would be right at home on Cruel Lucille"

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

>From [e-mail] (Amanda Walker)
Date: Fri, 27 May 1994 12:27:38 -0500
Organization: InterCon Systems Corporation, Herndon, VA USA

[e-mail] (Christopher Songer) writes:
> Thanks for the review -- it is much appreciated and was quite
> informative. Can you speak to the source control management provided
> by the CodeWarrior package?

Currently, it is Projector-aware, so we're continuing to use SourceServer from
MPW.  Code Warrior does not have its own source code control system.


Amanda Walker
InterCon Systems Corporation



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

>From [e-mail] (Nathan J. Wilson)
Subject: Refinement of SC++ 7.0 code gen bug
Date: 27 May 1994 06:34:38 GMT
Organization: UC Santa Cruz CIS/CE

First, I must say that I have been pleased with the response to my
original posting (Thanks Tom and Phil!).  I managed to reduce the
bug to a small stand alone example.  I have already send this to
Symantec directly, but I thought I should post it here so other's
could know what to watch out for until a fix is available.  The
causes appears to be with assign in the results of virtual
member function that returns a double directly to a member of
another class (or struct) within a loop.  The effect is that some
unpreserved registers get trashed resulting in garbage.  Work around
is to assign the result to a temporary.

-Nathan

// main.cpp

// Demonstration of a code generation bug in Symantec C++ 7.0
// Output should be:
// 12.34 == <garbage float>
//
// If you set BUG to 0 then you get the correct output which should be
// 12.34 == 12.34
//

#define BUG 1

#include <stdio.h>

#define THE_TYPE double // Must be double, float works fine

class FeatureNum {
public:
  THE_TYPE num; // Destination must be a member of a class or struct
};

class Foo {
public:
  virtual THE_TYPE Eval() { return 12.34; }; // Function must be a virtual member function
};

int main()
{
  Foo *foo = new Foo(); // Must be pointer to an object.  Can't use Foo foo;
  FeatureNum table2[1];
  long i = 0;
  while (i < 1) { // Must have loop
#if BUG
      table2[i].num = foo->Eval(); // Obviously, assignment must be direct
#else
      THE_TYPE tmp = foo->Eval();
      table2[i].num = tmp;
#endif
      printf("%lg == %lg\n", foo->Eval(), table2[i].num);
      i++;
  }
  delete foo;
  return 0;
}

#ifdef NEVER

Here is the disassembled buggy code:

main:
00000000: 4E56 FFDC          LINK      A6,#$FFDC
00000004: 48E7 1838          MOVEM.L   D3/D4/A2-A4,-(A7)
00000008: 7004               MOVEQ     #$04,D0
0000000A: 2F00               MOVE.L    D0,-(A7)
0000000C: 4EBA 0000          JSR       operator new(unsigned int)
00000010: 2840               MOVEA.L   D0,A4
00000012: 4A80               TST.L     D0
00000014: 670C               BEQ.S     *+$000E        ; 00000022
00000016: 2D40 FFE8          MOVE.L    D0,$FFE8(A6)
0000001A: 41ED 0000          LEA       Foo::_vtbl,A0
0000001E: 2240               MOVEA.L   D0,A1
00000020: 2288               MOVE.L    A0,(A1)
00000022: 200C               MOVE.L    A4,D0
00000024: 2640               MOVEA.L   D0,A3
00000026: 7000               MOVEQ     #$00,D0
00000028: 2800               MOVE.L    D0,D4
0000002A: 0C84 0000 0001     CMPI.L    #$00000001,D4
00000030: 6C00 0096          BGE       *+$0098        ; 000000C8
00000034: 2F0B               MOVE.L    A3,-(A7)
00000036: 204B               MOVEA.L   A3,A0
00000038: 2250               MOVEA.L   (A0),A1
0000003A: 2251               MOVEA.L   (A1),A1
0000003C: 4E91               JSR       (A1)
0000003E: 2604               MOVE.L    D4,D3
00000040: 2003               MOVE.L    D3,D0
00000042: C6FC 000A          MULU.W    #$000A,D3
00000046: 4840               SWAP      D0
00000048: C0FC 000A          MULU.W    #$000A,D0
0000004C: 4840               SWAP      D0
0000004E: 4240               CLR.W     D0
00000050: D680               ADD.L     D0,D3
00000052: 45EE FFDC          LEA       $FFDC(A6),A2
00000056: 3580 3000          MOVE.W    D0,$00(A2,D3.W)
0000005A: 2581 3002          MOVE.L    D1,$02(A2,D3.W)
0000005E: 2588 3006          MOVE.L    A0,$06(A2,D3.W)
00000062: 2004               MOVE.L    D4,D0
00000064: 2200               MOVE.L    D0,D1
00000066: C0FC 000A          MULU.W    #$000A,D0
0000006A: 4841               SWAP      D1
0000006C: C2FC 000A          MULU.W    #$000A,D1
00000070: 4841               SWAP      D1
00000072: 4241               CLR.W     D1
00000074: D081               ADD.L     D1,D0
00000076: 2D72 0000 FFF6     MOVE.L    $00(A2,D0.W),$FFF6(A6)
0000007C: 2D72 0004 FFFA     MOVE.L    $04(A2,D0.W),$FFFA(A6)
00000082: 3D72 0008 FFFE     MOVE.W    $08(A2,D0.W),$FFFE(A6)
00000088: 2F2E FFFC          MOVE.L    $FFFC(A6),-(A7)
0000008C: 2F2E FFF8          MOVE.L    $FFF8(A6),-(A7)
00000090: 3F2E FFF6          MOVE.W    $FFF6(A6),-(A7)
00000094: 2F0B               MOVE.L    A3,-(A7)
00000096: 204B               MOVEA.L   A3,A0
00000098: 2050               MOVEA.L   (A0),A0
0000009A: 2050               MOVEA.L   (A0),A0
0000009C: 4E90               JSR       (A0)
0000009E: 3D40 FFEC          MOVE.W    D0,$FFEC(A6)
000000A2: 2D41 FFEE          MOVE.L    D1,$FFEE(A6)
000000A6: 2D48 FFF2          MOVE.L    A0,$FFF2(A6)
000000AA: 2F2E FFF2          MOVE.L    $FFF2(A6),-(A7)
000000AE: 2F2E FFEE          MOVE.L    $FFEE(A6),-(A7)
000000B2: 3F2E FFEC          MOVE.W    $FFEC(A6),-(A7)
000000B6: 486D 0000          PEA       $0000(A5)
000000BA: 4EBA 0000          JSR       printf
000000BE: 4FEF 0018          LEA       $0018(A7),A7
000000C2: 5284               ADDQ.L    #$1,D4
000000C4: 6000 FF64          BRA       *-$009A        ; 0000002A
000000C8: 2F0B               MOVE.L    A3,-(A7)
000000CA: 4EBA 0000          JSR       operator delete(void *)
000000CE: 7000               MOVEQ     #$00,D0
000000D0: 4CDF 1C18          MOVEM.L   (A7)+,D3/D4/A2-A4
000000D4: 4E5E               UNLK      A6
000000D6: 4E75               RTS
000000D8

Foo::Eval(void):
00000000: 4E56 FFF4          LINK      A6,#$FFF4
00000004: 2D7C 4002 C570     MOVE.L    #$4002C570,$FFF6(A6)
          FFF6
0000000C: 2D7C A3D7 0A3D     MOVE.L    #$A3D70A3D,$FFFA(A6)
          FFFA
00000014: 3D7C 7000 FFFE     MOVE.W    #$7000,$FFFE(A6)
0000001A: 302E FFF6          MOVE.W    $FFF6(A6),D0
0000001E: 222E FFF8          MOVE.L    $FFF8(A6),D1
00000022: 206E FFFC          MOVEA.L   $FFFC(A6),A0
00000026: 4E5E               UNLK      A6
00000028: 225F               MOVEA.L   (A7)+,A1
0000002A: 584F               ADDQ.W    #$4,A7
0000002C: 4ED1               JMP       (A1)
0000002E

Here is the disassembled non-buggy code.  Changes are commented:

main:
00000000: 4E56 FFD0          LINK      A6,#$FFD0
00000004: 48E7 1838          MOVEM.L   D3/D4/A2-A4,-(A7)
00000008: 7004               MOVEQ     #$04,D0
0000000A: 2F00               MOVE.L    D0,-(A7)
0000000C: 4EBA 0000          JSR       operator new(unsigned int)
00000010: 2840               MOVEA.L   D0,A4
00000012: 4A80               TST.L     D0
00000014: 670C               BEQ.S     *+$000E        ; 00000022
00000016: 2D40 FFE8          MOVE.L    D0,$FFE8(A6)
0000001A: 41ED 0000          LEA       Foo::_vtbl,A0
0000001E: 2240               MOVEA.L   D0,A1
00000020: 2288               MOVE.L    A0,(A1)
00000022: 200C               MOVE.L    A4,D0
00000024: 2640               MOVEA.L   D0,A3
00000026: 7000               MOVEQ     #$00,D0
00000028: 2800               MOVE.L    D0,D4
0000002A: 0C84 0000 0001     CMPI.L    #$00000001,D4
00000030: 6C00 00A8          BGE       *+$00AA        ; 000000DA
00000034: 2F0B               MOVE.L    A3,-(A7)
00000036: 204B               MOVEA.L   A3,A0
00000038: 2250               MOVEA.L   (A0),A1
0000003A: 2251               MOVEA.L   (A1),A1
0000003C: 4E91               JSR       (A1)
0000003E: 3D40 FFDC          MOVE.W    D0,$FFDC(A6) ; Added
00000042: 2D41 FFDE          MOVE.L    D1,$FFDE(A6) ; Added
00000046: 2D48 FFE2          MOVE.L    A0,$FFE2(A6) ; Added
0000004A: 2604               MOVE.L    D4,D3
0000004C: 2003               MOVE.L    D3,D0
0000004E: C6FC 000A          MULU.W    #$000A,D3
00000052: 4840               SWAP      D0
00000054: C0FC 000A          MULU.W    #$000A,D0
00000058: 4840               SWAP      D0
0000005A: 4240               CLR.W     D0
0000005C: D680               ADD.L     D0,D3
0000005E: 45EE FFD0          LEA       $FFD0(A6),A2
00000062: 25AE FFDC 3000     MOVE.L    $FFDC(A6),$00(A2,D3.W) ; Change
00000068: 25AE FFE0 3004     MOVE.L    $FFE0(A6),$04(A2,D3.W) ; Change
0000006E: 35AE FFE4 3008     MOVE.W    $FFE4(A6),$08(A2,D3.W) ; Change
00000074: 2004               MOVE.L    D4,D0
00000076: 2200               MOVE.L    D0,D1
00000078: C0FC 000A          MULU.W    #$000A,D0
0000007C: 4841               SWAP      D1
0000007E: C2FC 000A          MULU.W    #$000A,D1
00000082: 4841               SWAP      D1
00000084: 4241               CLR.W     D1
00000086: D081               ADD.L     D1,D0
00000088: 2D72 0000 FFF6     MOVE.L    $00(A2,D0.W),$FFF6(A6)
0000008E: 2D72 0004 FFFA     MOVE.L    $04(A2,D0.W),$FFFA(A6)
00000094: 3D72 0008 FFFE     MOVE.W    $08(A2,D0.W),$FFFE(A6)
0000009A: 2F2E FFFC          MOVE.L    $FFFC(A6),-(A7)
0000009E: 2F2E FFF8          MOVE.L    $FFF8(A6),-(A7)
000000A2: 3F2E FFF6          MOVE.W    $FFF6(A6),-(A7)
000000A6: 2F0B               MOVE.L    A3,-(A7)
000000A8: 204B               MOVEA.L   A3,A0
000000AA: 2050               MOVEA.L   (A0),A0
000000AC: 2050               MOVEA.L   (A0),A0
000000AE: 4E90               JSR       (A0)
000000B0: 3D40 FFEC          MOVE.W    D0,$FFEC(A6)
000000B4: 2D41 FFEE          MOVE.L    D1,$FFEE(A6)
000000B8: 2D48 FFF2          MOVE.L    A0,$FFF2(A6)
000000BC: 2F2E FFF2          MOVE.L    $FFF2(A6),-(A7)
000000C0: 2F2E FFEE          MOVE.L    $FFEE(A6),-(A7)
000000C4: 3F2E FFEC          MOVE.W    $FFEC(A6),-(A7)
000000C8: 486D 0000          PEA       $0000(A5)
000000CC: 4EBA 0000          JSR       printf
000000D0: 4FEF 0018          LEA       $0018(A7),A7
000000D4: 5284               ADDQ.L    #$1,D4
000000D6: 6000 FF52          BRA       *-$00AC        ; 0000002A
000000DA: 2F0B               MOVE.L    A3,-(A7)
000000DC: 4EBA 0000          JSR       operator delete(void *)
000000E0: 7000               MOVEQ     #$00,D0
000000E2: 4CDF 1C18          MOVEM.L   (A7)+,D3/D4/A2-A4
000000E6: 4E5E               UNLK      A6
000000E8: 4E75               RTS
000000EA

Foo::Eval(void):
00000000: 4E56 FFF4          LINK      A6,#$FFF4
00000004: 2D7C 4002 C570     MOVE.L    #$4002C570,$FFF6(A6)
          FFF6
0000000C: 2D7C A3D7 0A3D     MOVE.L    #$A3D70A3D,$FFFA(A6)
          FFFA
00000014: 3D7C 7000 FFFE     MOVE.W    #$7000,$FFFE(A6)
0000001A: 302E FFF6          MOVE.W    $FFF6(A6),D0
0000001E: 222E FFF8          MOVE.L    $FFF8(A6),D1
00000022: 206E FFFC          MOVEA.L   $FFFC(A6),A0
00000026: 4E5E               UNLK      A6
00000028: 225F               MOVEA.L   (A7)+,A1
0000002A: 584F               ADDQ.W    #$4,A7
0000002C: 4ED1               JMP       (A1)
0000002E

Finally, here is a text version of the project file used to compile the code:
Wombat:Apps:ThinkC/C++:Play:Play.9
1 segments
3 files

Segment 3
Name:
Preload: false
Protected: true
Locked: true
Purgeable: true
SystemHeap: false
3 files:
Wombat:Apps:ThinkC/C++:SC++7.0:Standard Libraries:ANSI++
Wombat:Apps:ThinkC/C++:SC++7.0:Standard Libraries:CPlusLib
Wombat:Apps:ThinkC/C++:Play:main.cpp

#endif


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

>From [e-mail] (Fever)
Subject: Serial Port Programming?
Date: 26 May 1994 20:45:00 GMT
Organization: StarNet Communications, Inc

Could someone direct me to where I can find information on programming the
Mac's serial ports?  (Specifically: For modems, in C)

I'd appreciate any information...



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

>From [e-mail] (Jon Wdtte)
Date: 27 May 1994 06:48:46 GMT
Organization: The Royal Institute of Technology

In <2s31oc$[e-mail]> [e-mail] (Fever) writes:

>Could someone direct me to where I can find information on programming the
>Mac's serial ports?  (Specifically: For modems, in C)

The comp.sys.mac.programmer FAQ answer sheet shows, in code, how
to make a list of all available serial ports, and then go on to
use it for something. You can also check Think Reference 2.0.

nada.kth.se:pub/hacks/mac-faq

Cheers,

                    / h+

--
 -- Jon W{tte, [e-mail], Mac Software Engineer Deluxe --
  "After I first used the Mac, I had a dream in which I would walk up to
   people, touch them, and they'd turn black. Then I could talk to them."
                     -- Anon

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

>From [e-mail] (Eddy J. Gurney)
Date: Sat, 28 May 1994 00:49:06 GMT
Organization: Hewlett-Packard VCD

Fever ([e-mail]) wrote:
>Could someone direct me to where I can find information on programming the
>Mac's serial ports?  (Specifically: For modems, in C)

Check out the sample code on ftp.apple.com in:

/dts/mac/sc/snippets/devices/pbxxx-serial-demo.hqx

It is a good example of using the Mac serial ports "correctly".

--
Eddy J. Gurney N8FPW   Hewlett-Packard Company, Vancouver (USA!) Division
[e-mail]                       #include <standard-disclaimer.h>
"Failures are divided into two classes-- those who thought and never did,
      and those who did and never thought."     John Charles Salak

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

>From [e-mail] (Scott Kaplan)
Subject: Serial port use
Date: 24 May 1994 19:59:57 GMT
Organization: The Santa Fe Institute

Hi there.   I'm writing a quick little program which needs to make use of the
serial ports (specifically, the modem port attached to, of all things, a
modem...).
For what this program is doing, it does not seem worth my time to figure out
how to use the Comm Toolbox, so I thought I'd do it directly by opening the
serial drivers and using good ol' FSRead and FSWrite.  I'm having trouble,
and I was hoping someone could tell me why.

Here's the problem.  I've go to do an FSRead on the modem port.  If there is
something waiting in the modem port's buffer, everything's great.  However,
if there is nothing there (nothing has come in since I read from it last),
I die.  Rather than returning a count of 0 bytes read in, and rather than
returning some error value from FSRead, the machine crashes completely.  I
have to reboot.  So, I am clearly doing something wrong here...

If someone knows what this might be, please let me know.  If more information
about what I'm doing is needed, please ask, I'll be happy to provide it.  and
if you think that I shouldn't be avoiding the Comm Toolbox and that it would
easily solve my problems, let me know about that too.

Thanks muchly.

Scott Kaplan


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

>From [e-mail] (Eddy J. Gurney)
Date: Tue, 24 May 1994 23:23:40 GMT
Organization: Hewlett-Packard VCD

Scott Kaplan ([e-mail]) wrote:
>Hi there.   I'm writing a quick little program which needs to make use of the
>serial ports (specifically, the modem port attached to, of all things, a
>modem...).
>For what this program is doing, it does not seem worth my time to figure out
>how to use the Comm Toolbox, so I thought I'd do it directly by opening the
>serial drivers and using good ol' FSRead and FSWrite.  I'm having trouble,
>and I was hoping someone could tell me why.

Check out the sample code on ftp.apple.com:

/dts/mac/sc/snippets/devices/pbxxx-serial-demo.hqx

The snippets directory is a GREAT place to look for answers to ALL of
your questions BEFORE posting to net...

--
Eddy J. Gurney N8FPW   Hewlett-Packard Company, Vancouver (USA!) Division
[e-mail]                       #include <standard-disclaimer.h>
"Failures are divided into two classes-- those who thought and never did,
      and those who did and never thought."     John Charles Salak

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

>From [e-mail] (Andrew Bradley)
Date: Thu, 26 May 1994 09:03:15 GMT
Organization: Crosfield, Hemel Hempstead, UK

In article <2rtmbt$[e-mail]> [e-mail] (Scott Kaplan) writes:
>Hi there.   I'm writing a quick little program which needs to make use of the
>serial ports (specifically, the modem port attached to, of all things, a
>modem...).

>Here's the problem.  I've go to do an FSRead on the modem port.  If there is
>something waiting in the modem port's buffer, everything's great.  However,
>if there is nothing there (nothing has come in since I read from it last),
>I die.  Rather than returning a count of 0 bytes read in, and rather than
>returning some error value from FSRead, the machine crashes completely.  I
>have to reboot.  So, I am clearly doing something wrong here...
>


I use SerGetBuf to check whats available and read that in - if its zero
I use my own minimum data size because otherwise the FSRead will hang
until it got all the bytes you asked for.

Remember the Mac may be written in C but its not UNIX. You have to study
the Mac libraries very carefully and forget all your unix past !

Regards Andrew



--


  |\/\/\/\                   Andrew Bradley
  |      |                   Crosfield Electronics Limited

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

>From [e-mail] (Chris Swan)
Date: Thu, 26 May 1994 21:25:04 GMT
Organization: Actrix Information Exchange

Scott Kaplan <[e-mail]> wrote:
> Hi there.   I'm writing a quick little program which needs to make use of the
> serial ports (specifically, the modem port attached to, of all things, a
> modem...).
> Here's the problem.  I've go to do an FSRead on the modem port.  If there is
> something waiting in the modem port's buffer, everything's great.  However,
> if there is nothing there (nothing has come in since I read from it last),
> I die.  Rather than returning a count of 0 bytes read in, and rather than
> returning some error value from FSRead, the machine crashes completely.  I
> have to reboot.  So, I am clearly doing something wrong here...

I'm not sure why you're crashing but the normal way is to call:

theError:= SerGetBuf(refNum,count);
if (theError=noErr) and (count>0) then
  theError:= FSRead(refNum,count,buffPtr);

works for me !
--
Chris Swan
[e-mail]
--


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

>From [e-mail] (Bruce Hoult)
Date: Fri, 27 May 1994 14:08:43 +1200 (NZST)
Organization: (none)

[e-mail] (Scott Kaplan) writes:
> Here's the problem.  I've go to do an FSRead on the modem port.  If there is
> something waiting in the modem port's buffer, everything's great.  However,
> if there is nothing there (nothing has come in since I read from it last),
> I die.  Rather than returning a count of 0 bytes read in, and rather than
> returning some error value from FSRead, the machine crashes completely.  I
> have to reboot.  So, I am clearly doing something wrong here...

The machine hasn't crashed -- it's just waiting for the data you asked for.

If you want your program to continue if there's nothing there then use...

   SerGetBuf(refNum:INTEGER;VAR count:LONGINT):OSErr;

...to find out how many bytes are available, and then ask for at most
that many in your FSRead.

-- Bruce

btw, I just used ObiWan to paste the definition of SerGetBuf.  It's nice, but
what I *hate* about it is that the first time I use it after a restart, it
puts up a much larger window and all my Icons that are nicely arranged along
the bottom of my Finder desktop somehow get moved to the RH side of the desktop.

::aaaarggghhhh::

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

>From [e-mail] (Pete Gontier)
Date: Fri, 27 May 1994 21:33:37 GMT
Organization: cellular

[e-mail] (Andrew Bradley) writes:

>Remember the Mac may be written in C but its not UNIX.

I know it's a persistent rumor, but the Mac ROMs were not written in any
particular high-level language. I guess from now on the Power Mac ROMs
will mostly be written in C, or perhaps C++, but from what I understand,
much of that work has not yet been released. It's still mostly 68K
assembly using the Pascal calling convention.
--
 Pete Gontier, CTO, Integer Poet Software; [e-mail]

 "Bedrock does support a form of drag and drop. Development dragged on
 for about two years, and then was dropped." -- Brian Clark

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

>From [e-mail] (Scott Kaplan)
Date: 29 May 1994 16:43:44 GMT
Organization: The Santa Fe Institute

Many thanks to those that responded!  I know now of a very useful place
for sample source code from Apple, which will always be a good thing.  More-
over, I now know what I was doing wrong.  Sure enough, the use of SerGetBuf
fixed my troubles.

My apologies for not having read through the serial driver toolbox routines
more carefully.  It's been a little while since I had done Mac programming,
so I had fallen out of that good habit.

Scott Kaplan
[e-mail]


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

>From [e-mail] (Steven G. Johnson)
Subject: THINK Class Library FTP Site?
Date: 26 May 1994 18:32:48 GMT
Organization: Massachvsetts Institvte of Technology

There used to be an FTP site containing an archive of classes that
people have written for TCL.  It was connected to a mailing list
(called TCL-talk, I think), and was located at ftp.brown.edu.
However, I looked at this site today and I couldn't find anything!
Does anyone know what has happened to the archive (and the mailing
list)?  Has it been moved to someplace else?  If so, where?

Please email all responses, as it is difficult for me to read this
newsgroup at the moment.  I will post a summary of the replies if
anyone is interested.  Thanks!

                    Cordially,
                    Steven G. Johnson

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

>From [e-mail] (Steven G. Johnson)
Date: 27 May 1994 16:28:35 GMT
Organization: Massachvsetts Institvte of Technology

I recently posted a request for the location of the FTP site
containing TCL classes and functions.  I was told that the new site
is:    daemon.ncsa.uiuc.edu  (I checked it out and this is indeed the
correct address).

Thanks to all who replied.

                        Cordially,
                        Steven G. Johnson

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

>From Will Husain <[e-mail]>
Subject: storage for XCMDS
Date: 23 May 1994 14:51:52 GMT
Organization: k12

I'd like to write an XCMD for hypercard that will store information
between successive calls from hypercard.  How can I keep track of global
data with this kind of resource?

Thanks for any suggestions

Will Glass-Husain
Catalina Foothills High School
[e-mail]

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

>From [e-mail] (Alex Metcalf)
Date: Tue, 24 May 1994 12:02:21 GMT
Organization: Best Before Yesterday

In article <2rqfu8$[e-mail]>, Will Husain
<[e-mail]> wrote:

> I'd like to write an XCMD for hypercard that will store information
> between successive calls from hypercard.  How can I keep track of global
> data with this kind of resource?

You first need to make up your own struct of information which you want to
keep, something like this:

typedef struct
{
   Ptr    myMemory;
   long   myLong;
   short  myShort;
} MyData, *MyDataPtr;

So, you use NewPtr to make some space on the heap for your data. Then, you
take the pointer (to your data), typecast it to a 'long', use NumToString,
and store the string (which is actually the address of your info) in a
HyperCard global variable using SetGlobal.

Next time round, use GetGlobal, use StringToNum, and then typecast back to
a pointer. You can then access your information.

Hope this helps! Get in touch if you have any other questions.


Best wishes,




Alex

--
Alex Metcalf, Best Before Yesterday
Mac programmer in C, C++, HyperTalk, assembler

Internet, AOL, BIX: [e-mail]            "Surely you
AppleLink:          [e-mail]@internet#   can't be
CompuServe:         INTERNET:[e-mail]    serious?"
Delphi:             [e-mail]@inet#
FirstClass:         [e-mail],Internet   "I am serious...
Fax (UK):           (0570) 45636                         and don't call
Fax (US / Canada):  011 44 570 45636                     me Shirley."

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

>From [e-mail] (Arne Venstad)
Date: 25 May 1994 08:44:02 GMT
Organization: SINTEF DELAB

In article <[e-mail]>,
[e-mail] (Alex Metcalf) wrote:
>
> In article <2rqfu8$[e-mail]>, Will Husain
> <[e-mail]> wrote:
>
> > I'd like to write an XCMD for hypercard that will store information
> > between successive calls from hypercard.  How can I keep track of global
> > data with this kind of resource?
>
> You first need to make up your own struct of information which you want to
> keep, something like this:
>
> typedef struct
> {
>    Ptr    myMemory;
>    long   myLong;
>    short  myShort;
> } MyData, *MyDataPtr;
>
> So, you use NewPtr to make some space on the heap for your data. Then, you
> take the pointer (to your data), typecast it to a 'long', use NumToString,
> and store the string (which is actually the address of your info) in a
> HyperCard global variable using SetGlobal.
>
>...

I do not think it is wise to allocate locked memory blocks behind
the back of HyperCard. I would therefore suggest you use NewHandle
instead of NewPtr, and always unlock it before returning to HyperCard.
Otherwise the schema is ok.

Arne Venstad    Internet: [e-mail]
SINTEF DELAB,
N-7034 Trondheim,
Norway.

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

>From [e-mail] (Alex Metcalf)
Date: Wed, 25 May 1994 12:01:50 GMT
Organization: Best Before Yesterday

In article <[e-mail]>,
[e-mail] (Arne Venstad) wrote:

> In article <[e-mail]>,
> [e-mail] (Alex Metcalf) wrote:
> >
> > In article <2rqfu8$[e-mail]>, Will Husain
> > <[e-mail]> wrote:
> >
> > > I'd like to write an XCMD for hypercard that will store information
> > > between successive calls from hypercard.  How can I keep track of global
> > > data with this kind of resource?
> >
> > You first need to make up your own struct of information which you want to
> > keep, something like this:
> >
> > typedef struct
> > {
> >    Ptr    myMemory;
> >    long   myLong;
> >    short  myShort;
> > } MyData, *MyDataPtr;
> >
> > So, you use NewPtr to make some space on the heap for your data. Then, you
> > take the pointer (to your data), typecast it to a 'long', use NumToString,
> > and store the string (which is actually the address of your info) in a
> > HyperCard global variable using SetGlobal.
> >
> >...
>
> I do not think it is wise to allocate locked memory blocks behind
> the back of HyperCard. I would therefore suggest you use NewHandle
> instead of NewPtr, and always unlock it before returning to HyperCard.
> Otherwise the schema is ok.

I've had problems before with NewHandle: HyperCard enjoys moving memory
around, especially when stacks are switched/closed. If HyperCard (or the
system memory fuctions) decide to purge your block then there's no way to
get your data back.



Alex

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

>From [e-mail] (Ron Hunsinger)
Date: Wed, 25 May 94 22:52:20 PST
Organization: Berkeley Macintosh Users Group

Will Husain <[e-mail]> writes:

>I'd like to write an XCMD for hypercard that will store information
>between successive calls from hypercard.  How can I keep track of global
>data with this kind of resource?

Store the data in a relocatable block, and convert the Handle to that
block to a string and store the string in a global variable.  (Choose a
name that won't be used by somebody else.)

On entry to the XCMD, get the value of that variable.  If it doesn't
have a value, you know this is the first call of the XCMD, and you should
allocate your data.  If the variable does have a value, convert it
back to a long, and cast it as a Handle to your data.

-Ron Hunsinger

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

>From Kevin Marks <[e-mail]>
Date: Sat, 28 May 1994 11:08:05 GMT
Organization: Me

In article <[e-mail]> Ron Hunsinger, [e-mail]
writes:
>Store the data in a relocatable block, and convert the Handle to that
>block to a string and store the string in a global variable.  (Choose a
>name that won't be used by somebody else.)

This is a bit dodgy - you can't guarantee that someone else won't trash
your variable...
The best way is to write a XWindow instead, and store your Handle in the
window's refcon. You don't have to show the window if it's not working on
visual stuf, and thsi means you can send messages to it and set
properties, so your HyperTalk looks less obscure.

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

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


ˇ