Re: About a mass bug report not based on Sid or Jessie.
On Wed, 16 Apr 2014, Charles Plessy wrote:
> I invite people interested in new ports to work on that change: run autoreconf
> by default. The other approach, to run autoreconf only when needed, is
> never-ending: new packages uploaded today build fine everywhere, but some of
> them will need patching in the future. Please break the circle !
Hmm. "The other approach". You seem to imply that there are only two
possible approaches here. We should probably aim for a solution which
is better than the just two we are considering here.
I have also received one of those bugs asking for dh-autoreconf to be
put in debian/rules (for wdiff). I forwarded it upstream, and a new
upstream release was made in a matter of days.
There are a number of reasons why I dislike this dh-autoreconf thing.
IMHO, there is something fundamentally wrong when we have to copy the
same file over and over again so that builds do not fail. The GNU
system was designed to be portable, and yet the GNU project is
shipping source tar.gz files that do not work on newly created
architectures.
I would call that a serious design flaw.
What some people here try to do, update config.guess and related files,
is, IMHO, just a hack. Sure, it will just work, but only for us (Debian).
Other distributors will still have the same problem.
If it is so much important that the right file is used, why is the
file not part of gcc, libc or some essential package as a "system file"?
Can we think about a solution for this instead of *just* try to make
it work by using a hack?
Anybody from the GNU project reading this?
BTW: I know very well how much work automatic dh-autoreconf would
save. I still remember the 200 source-only NMUs (followed by binary
uploads) to "unreleased" that I did for kfreebsd-* at ftp.gnuab.org a
long time ago because of outdated config.{guess,sub} files. But I
still think this is the wrong solution to the problem in the long
term.
Thanks.
Reply to: