On Tue, 28 Aug 2007 21:15:35 +0300, Riku Voipio <[EMAIL PROTECTED]> said:
>> >> Many packages FTBFS (silently!) if an environment variable TAPE is
>> >> set.
>> Perhaps dpkg-buildpackage should unset TAPE...?
> pbuilder and other tools already do that when chrooting? Tar's $TAPE
> behaviour fails the principle of least suprise. Tar developers should
> reconsider the usability implications of such feature.
> For packages, I think $TAPE is hardly the only environment variable
> that can change package outcome. Atleast the locale variables can
> wreck havoc for builds. Or consider setting AWKPATH or even PATH to
> something strange... Clearly packagers should not need to take care of
> all possible enviroment misconfigurations.
> That's why we tell people to use pbuilder.
I think I disagree with the reason given for this advice. What
is the end goal that we are trying to achieve? Is it to upload binary
packages that build despite leaving flaws i the build process? Always
building in pbuilder masks errors like the $TAPE error; we would have
been better off having the build fail and be corrected.
Are we not supposed to be a valuable members of the free
software community? Part of the promise of free software is that
people *SHARE* source code; so our sources should be robust enough for
our users to tinker with on their machines, and to be buildable on
development platforms, not just some sanitized hothouse build
environment.
I posit that always building in make software more fragile (and
more standardized, and repeatable, which is good, but fragile
nevertheless).
Me, I would try to build in pbuilder while I am trying to figure
out what the build dependencies are, and periodically on new upstream
to ensure the build dependencies have not shifted, but other times I
would build and test on my own machine, to more closely replicate the
environment that downstream users might have.
manoj
--
Documentation is the castor oil of programming. Managers know it must be
good because the programmers hate it so much.
Manoj Srivastava <[EMAIL PROTECTED]> <http://www.debian.org/~srivasta/>
1024D/BF24424C print 4966 F272 D093 B493 410B 924B 21BA DABB BF24 424C
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
