From:	khan AT xraylith DOT wisc DOT edu (Mumit Khan)
Subject:	Re: popen and mingw32
9 Jul 1998 11:07:42 -0700 :
Message-ID:	<Pine.SUN.3.93.980709125132.14596D-100000.cygnus.cygwin32.developers@modi.xraylith.wisc.edu>
References:	<199807091702 DOT NAA14900 AT subrogation DOT cygnus DOT com>
Mime-Version:	1.0
To:	Ian Lance Taylor <ian AT cygnus DOT com>
Cc:	cygwin32-developers AT cygnus DOT com

On Thu, 9 Jul 1998, Ian Lance Taylor wrote:

> I've had a bug report about windres, running on Windows 95, using
> Mumit's gcc binaries which I believe are mingw32 based.
> 
> windres works by using popen to run `gcc -E'.  Based on the bug
> report, the bug appears to be that the gcc -E output is sent directly
> to the console, rather than being read by windres via popen.  windres
> then fails because it doesn't see any input.

I've gone thru all the variations and this is what I found:
  
  1. popen doesn't work on *some* of the Win95 machines using CRTDLL. 
     Does work however on my laptop, which is running Win95 Rev B with
     lots of the other software (some of which replace the system runtime
     DLLs).
  
  2. popen doesn't work on *some* of the NT machines using MSVCRT.DLL or
     MSVCRT40.DLL, but does work using MSVCRT20.DLL.

Seems to be quite sensitive to the versions. Now, it's interesting that
all the test code I ran worked perfectly using MSVC 4/5 using MSVCRT40.DLL, 
which means we're perhaps missing something in the startup code that 
tweaks who knows what.

> 
> I don't know if there is a real bug here, although I see that it is at
> least possible.  I don't have a way to test it.  I'm just passing this
> bug report along.  I plan to remove it from my mail file and not worry
> about it.
> 

I guess at some point I'll try a replacement popen to see if it'll fix it
or not. I myself had never noticed it since I cross-build everything, and
Douglas sent me the first report of this.

Mumit

- Raw text -