[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: configure and $TEMP
From: |
Mark D. Baushke |
Subject: |
Re: configure and $TEMP |
Date: |
Mon, 24 Nov 2003 11:56:30 -0800 |
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Derek Robert Price <derek@ximbiot.com> writes:
> Chris Lightfoot wrote:
>
> >On Sat, Nov 22, 2003 at 10:38:11PM -0500, Larry Jones wrote:
> >
> >
> >(The wider point is that I think saving the value of the
> >environment variable used at build time is really broken,
> >but obviously that's a matter of opinion....)
>
>
> That depends on whether you assume the builder had a good reason for
> setting the variable in the environment, but I can look at this your way
> too since the CVS executable itself will look at $TMPDIR, $TMP, & then
> $TEMP itself at runtime before it defaults to the value selected by
> configure. Anyone else care enough about this to register an opinion?
It makes sense to honor $TMPDIR during the build of the cvs executable,
just as it makes sense to honor the $CC environment variable.
However, I tend to agree with Chris Lightfoot, if an administrator wants
to set the default for the cvs executable to something, then it is best
to be explicit with the --with-tmpdir=$TMPDIR option.
If that change is adopted, then, to be consistent, the behavior of the
- --with-editor and the $EDITOR variable should also require and explicit
- --with-editor be provided. I have no problem if $EDITOR is used if none
of the default list of editors can be found (vim, vi, emacs, nano, pico,
edit).
I will note that the --with-umask option currently does NOT pay
attention to the umask of the user that is configuring cvs, so that
might be the lever to change the defaults for the others to also require
explicit statement in the configure step.
-- Mark
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (FreeBSD)
iD8DBQE/wmJu3x41pRYZE/gRAlPnAKC+joF8D/5mO3ReZzU3ACg23saOPgCfd7ww
rzTTDxwEkI79iWXSSFhsHU4=
=deoC
-----END PGP SIGNATURE-----