Mail Archives: djgpp/1998/05/11/12:46:50
Date sent: Mon, 11 May 1998 03:58:38 +0000 ( )
From: "Gurunandan R. Bhat" <grbhat AT unigoa DOT ernet DOT in>
Subject: Re: / and \ in gcc2.8.1
>
> On Mon, 11 May 1998, Andris Pavenis wrote:
>
> > > I find that a TMPDIR variable set to (a ramdisk) d:\ causes gcc
> > > to Abort! without any further messages (even with -v). Changing it to d:/
> > > removes this problem. gcc2.7.2.1 worked with d:\
> > >
> >
> > This is true only for is TMPDIR is set to root directory of drive (not necesarily
> > ramdisk) and backslash is used ('TMPDIR=d:\tmp' works Ok).
>
> May I add two points:
>
> 1) The same problem occurs even if the backslash is omitted completely. I
> mean:
>
> TMPDIR=d:
> in AUTOEXEC.BAT also causes gcc to silently "Abort!"
gcc tries to look if directory d: exist
access("D:",R_OK+W_OK+X_OK)
and access fails. Therefore gcc looks for environment parameters TMP
and TEMP and You perhaps have variable TMP which point to
D:\ (or root directory of some other drive) so the result does not change.
A temporary workaround for this problem could be specifying
TMPDIR=D:/
>
> 2) The -v option gives no indication that TMPDIR is invalid
> (something that happens with 2.7.2.1) since it merely prints out the
> location of the specs and dies.
>
If You want to test fixed (I hope) executables You can get them (only
gcc.exe, gxx.exe and cpp.exe) from URL
http://www.lanet.lv/~pavenis/gccfix.zip
Please replace executables from gcc281b.zip with ones from this archive.
Don't forget to copy new ccp.exe into $DJDIR/lib/lib-gcc/djgpp/2.81 but NOT
in $DJDIR/bin
Andris
- Raw text -