Re: [Reproducible-builds] ftbfs_build-indep_not_build_on_armhf, and add bochs to it

2016-06-04 Thread Holger Levsen
Hi Santiago,

thanks for bringing this up…!


On Fri, Jun 03, 2016 at 09:13:29PM +, Santiago Vila wrote:
> > +- ftbfs_build-indep_not_build_on_armhf
[...] 
> Sometimes, packages generating "Arch: all" binary packages have good
> reasons to require that those packages are built only under certain
> architectures.

sure. (though I'm not convinced failing to debuild is a good outcome,
though not failing and building different things on different archs is
also not a good outcome…)

> In the case of "bochs", there are some hints about the reasons in
> Bug #481147 (which I closed recently because, well, it seemed "as
> fixed as it can be" to me).
> 
> A similar case also happens with the "aboot" package, which generates an
> "Arch: all" package which apparently may only be generated under the
> alpha architecture (see Bug #805988).
> 
> 
> An "issue" suggests to me "something which has eventually to be fixed",
> but frankly, I don't think we should really require that those
> packages generate their "Arch: all" binary packages from any other
> architecture.
 
I think I shall improve the issue description to make it very clear that
this is just a meta-issue used to track that we currently get expected
FTBFS failures.

> So, instead of "this package needs to be fixed", those packages would
> maybe deserve a "this package should not be built on such architecture
> because it is simply not supposed to work".

the point of the issue was precicely to remove the "ftbfs in
reproducible builds tests" from tracker.d.o, so the maintainers arent
told there is something wrong.

> Do you think it would be possible to achieve the same result with a
> "banned packages" list which is architecture-specific instead of this
> funny issue?

yes, we could probably create another category of not-for-us or such,
but that will need quite some code changes, while adding this new issue
was very easy to do.

> (Or maybe your plan was to make the autobuilder to be aware of packages
> having this issue precisely to avoid the build?)

yeah, maybe they shouldnt be marked "ftbfs" on our pages neither. but
then its the correct result for how we build…


btw, yes, we could also rename this issue to
ftbfs_build-indep_not_build_some_archs or add new issues
ftbfs_build-indep_not_build_on_amd64 and
ftbfs_build-indep_not_build_on_i386, I'm undecided what's best.

-- 
cheers,
Holger


signature.asc
Description: Digital signature
___
Reproducible-builds mailing list
Reproducible-builds@lists.alioth.debian.org
http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/reproducible-builds

Re: [Reproducible-builds] ftbfs_build-indep_not_build_on_armhf, and add bochs to it

2016-06-04 Thread Mattia Rizzolo
On Fri, Jun 03, 2016 at 09:13:29PM +, Santiago Vila wrote:
> On Fri, Jun 03, 2016 at 04:49:10PM +, Holger Levsen wrote:
> > +ftbfs_build-indep_not_build_on_armhf:
> > +  description: |
> > +- ftbfs_build-indep_not_build_on_armhf

probably this tag name should lose the 'armhf' word in it, anyway.

> Sometimes, packages generating "Arch: all" binary packages have good
> reasons to require that those packages are built only under certain
> architectures.

yes, it's an annoying thing, but it's true.  I'd like to see
https://bugs.launchpad.net/launchpad/+bug/217427 for that.

> An "issue" suggests to me "something which has eventually to be fixed",
> but frankly, I don't think we should really require that those
> packages generate their "Arch: all" binary packages from any other
> architecture.

that was not the idea of this issue at all.  It's called "issue", but
it's better read as "tag" in these occasion.

> So, instead of "this package needs to be fixed", those packages would
> maybe deserve a "this package should not be built on such architecture
> because it is simply not supposed to work".

This issue, coupled with a change in jenkins code, makes possible for us
to not export the failure notice towards tracker.d.o/DDPO, so
maintainers should not be bothered with the "FTBFS" notice.
https://anonscm.debian.org/git/qa/jenkins.debian.net.git/commit/?id=f5c5274cea8cc4d7b58e374c7114f01c07019035

> Do you think it would be possible to achieve the same result with a
> "banned packages" list which is architecture-specific instead of this
> funny issue?

nobody likes blacklisting packages :(

-- 
regards,
Mattia Rizzolo

GPG Key: 66AE 2B4A FCCF 3F52 DA18  4D18 4B04 3FCD B944 4540  .''`.
more about me:  https://mapreri.org : :'  :
Launchpad user: https://launchpad.net/~mapreri  `. `'`
Debian QA page: https://qa.debian.org/developer.php?login=mattia  `-


signature.asc
Description: PGP signature
___
Reproducible-builds mailing list
Reproducible-builds@lists.alioth.debian.org
http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/reproducible-builds