It seems like a number of pieces are finally coming together for OS/2:
1. Some work is being done on OS2LDR and the kernel.
2. Openwatcom is getting wider use.
3. J. deB Pollard has a number of 32-bit replacements for 16-bit OS/2 code.
4. etc.

Is it possible, despite the lack of motion by the Voyager, OSFREE, and
FreePM projects, and the inaction by IBM on the open-source petition,
that eventally an OS/2 replacement will be available?

I've just come to the realization that emitting any debug info in
OpenWatcom disables all optimizations, regardless of the other
parameters. As a side-effect of rolling my own makefiles (rather than
using the ones generated automatically from the GUI), I re-evaluated
this and created a "debug" target and a "production" target (which
doesn't include debug info naturally).

Problem is, with nearly any kind of optimization I try, I get a
resulting application which crashes in a DLL that I didn't write. This
DLL was written in Sybil I believe, and I have the functions declared as:

extern "C"
{
ULONG __pascal functionName1( void );
ULONG __pascal functionName2( ULONG, LONG, LONG, LONG, LONG, ULONG );
HWND __pascal functionName3( ULONG );
...
}

I create an import library from the DLL using wlib:
wlib -n -b importlib.imp thedll.dll

On Sat, 8 Dec 2007 21:27:54 UTC, "Peter Weilbacher" wrote:

Just for the record: it seems that GpiCharString has all the logic for
"font shaping" built in, Mozilla never contained language specific code
for this. I wish there was a way I could access the tables that it uses
to do that...

I'm currently changing most of my programs to use High Memory by
using OBJ_ANY for DosAllocMem. So far on eCS this works fine and
I can allocate more than 512MB just fine.

I could not find any hint in OS2TK about backward compatibility.
Will DosAllocMem on older OS ignore OBJ_ANY and allocate from low
memory pool as fallback instead (pre Warp 4.5x and Warp 3)?

If it is not backward compatible, what is the recommended way
to detect the support?

[This was supposed to go to comp.os.os2.announce, but there seems to
be a problem getting messages posted there.]

An ALPHA version of the new Iron Spring PL/I compiler for OS/2 is now
available at:
http://www.iron-spring.com

This is a fairly rough alpha. It works well enough to compile itself
and its run-time library, but is missing many features not immediately
required to achieve a working compiler. The "release notes" included
in the package document some, but not all, of the current
restrictions. The 1.0 release is expected to be roughly a "Subset G"
compiler. Eventually the compiler will handle the full PL/I language
plus extensions from other compilers.

Following the OS/2 version a Linux version is planned.

OS/2 programmers familiar with PL/I might be interested in
experimenting with the current version. Subsequent major versions
will be announced as they become available.

Hi,

1.) DosQuerySysState is badly and insufficiently documented in the
"Programming Guide and Reference Addendum" (addendum.inf) that comes
with the OS/2 2.45 toolkit. But especially querying the module version
info (QS_MODVER) is NOT described there and works differently from what
the code example implies. It was debugging, trial and error.

2.) One additional thing I forgot to mention: you might get several hits
for a specific PID. That's because all DLLs loaded by a process are also
listed with the PID of the process that loaded them (first ?).
Unfortunately there is no easy way out of this dilemma. However my
observation is that EXEs are listed FIRST and then the DLLs follow. So
the first hit should be the right one when you are looking for EXEs.

Does OS/2 support copy on write memory? eg discussion on FFmpeg-dev
about large tables and whether all operating systems support one copy of
the table until a process writes to it.
Dave

I know I've done this before, but I can't remember how at this point. I
have a crash address from a user in DOSCALL1.DLL (0002:00009e3a is the
exact address). How do I find what procedure it was in (or at least get
a possible API call name, even if this particular function wasn't exported)?

Do I have to ask the user to set up a PROCDUMP for me? I have full
symbollic information and an OW map file for the executable.

A user running a piece of software that I wrote reports that he can't
get up a certain dialog window with the current software version (but
somehow past versions were fine). I added one new kind of control to
this dialog window - a standard pushbutton with a mini-icon picture
instead of text. The resource definition is as follows:

CONTROL "#2", PB_widgetID, , , , , WC_BUTTON,
BS_PUSHBUTTON | BS_MINIICON | WS_VISIBLE | WS_GROUP | WS_TABSTOP

This resource resides in a DLL, which also has an icon in the "#2" slot.
The executable itself only has a single resource of an icon in the
"#1" slot. Debug code that I injected shows some stuff going on as the
dialog window tries to initialize itself (I had it dump all window
procedure message traffic to a log). Ultimately it gets sent a
WM_DESTROY message before the WinLoadDlg function even returns. It
looks like this:

[TID01]: WinDefDlgProc( 2147484184, 83, 643300, 0 )
[TID01]: WinDefDlgProc( 2147484184, 77, 640896, 0 )
[TID01]: WinDefDlgProc( 2147484184, 86, 480765580, 480765464 )
[TID01]: WinDefDlgProc( 2147484184...