[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: FTBFS if built twice in a row

On 29/06/07 at 20:21 +0300, Gerard Lledó wrote:
> On 6/29/07, Santiago Vila <sanvila@unex.es> wrote:
> >On Fri, 29 Jun 2007, Adeodato Simó wrote:
> >
> >> * Santiago Vila [Fri, 29 Jun 2007 16:58:13 +0200]:
> >> > I don't know how much typical this example is, but the two "can't build
> >> > twice in a row" bugs I got were like that.
> >>
> >> The you should/could remove the .mo files in your debian/rules clean
> >> target.
> >
> >Sure. I did that.
> >
> >I was trying to explain that a complete "./configure; make; make clean"
> >cycle is really required to discover these kind of bugs, as opposed to the
> >"simplified check" that was being proposed.
> >
> >
> The "simplified check" that I proposed was to compare the sources
> before and after the debuild build / debuild clean process, not just
> at the end of it. Just one build is necessary. This could be
> implemented in debuild (or somewhere else, someone will know better
> than me) and prompt the maintainer like lintian does.
> Besides, (correct me if I'm wrong) I think the cases that you're
> proposing to break my ideal "clean" status are not usual. Of course,
> the timestamps/permissions can be skipped in the check in order to
> avoid a lot of packages to fail, but a warning at the end of the build
> could indicate to the mantainer that some files are left "uncleaned"
> or if some have unexpectedly changed.
> In any case, feel free to ignore me if my ideas are completely wrong ^^.

The problem is that it isn't required to have exactly the same source
tree after "./configure ; make ; make clean". It's just required that
"./configure ; make ; make clean ; ./configure ; make" works. It is
possible that the first build modifies some files, but that the package
can still be built, without being differerent from the one from the
first build.

However, what was not tested yet, is whether the packages generated
during the first and second build are identical. This second step should
probably be done when most of the bugs of the first step will have been
| Lucas Nussbaum
| lucas@lucas-nussbaum.net   http://www.lucas-nussbaum.net/ |
| jabber: lucas@nussbaum.fr             GPG: 1024D/023B3F4F |

Reply to: