Message-ID:	<35435F39.66E2BD75@gmx.net>
Date:	Sun, 26 Apr 1998 16:22:17 +0000
From:	Robert Hoehne <robert DOT hoehne AT gmx DOT net>
Organization:	none provided
MIME-Version:	1.0
To:	Andris Pavenis <pavenis AT lanet DOT lv>
CC:	djgpp-workers AT delorie DOT com, Vik Heyndrickx <Vik DOT Heyndrickx AT rug DOT ac DOT be>
Subject:	Re: Generation of temporary files for gcc-2.8.1
References:	<199804241652 DOT SAA34998 AT ieva06 DOT lanet DOT lv>

Andris Pavenis wrote :
> 
> ready to be uploaded. The only question is whether  we can still provide
> such sharing in final release of the gcc port.

At first I would say here, that this feature is not so much
important to depend the other working gcc on it. Currently
I solve the problem (if I really need it) by simply defining
an other TMPDIR when running make, for instance

make TMPDIR=. ....

> Use of some artificially generated PID is perhaps not quick solution as requires
> rather deep study for various systems and even then it's rather likely that
> danger of conflicts between 2 copies of gcc will still remain.

My first thought about using the cs selector is wrong like I tested
yesterday, since it will the same in different DOS boxes, but now I
found, that DJGPP provides also a function getpid() which reads out
(when I saw it correct) the BIOS area of the current time. So this
value could be used in the mktemp-template.

>    - I'm removing definition of MKTEMP_EACH_FILE

With the above hint, this is not needed.

>    - I'm replacing libc.a version of mktemp() (for gcc.exe and gxx.exe only)

In general I'm not a friend of such hacks, since it might cause
problems in the future (like you mentioned it already).

Robert
-- 
******************************************************
* email:   Robert Hoehne <robert DOT hoehne AT gmx DOT net>     *
* Post:    Am Berg 3, D-09573 Dittmannsdorf, Germany *
* WWW:     http://www.tu-chemnitz.de/~sho/rho        *
******************************************************

- Raw text -