Bug#818990: Is it really a flint problem?
On Sat, May 21, 2016 at 08:47:06AM +0200, Julien Puydt wrote: > Hi, > > On 21/05/2016 01:46, Santiago Vila wrote: > > It still fails on reproducible builds on the "armhf" architecture, but > > only in testing, and the way it fails is now completely different than > > the way it failed on amd64 in the initial report. > > But it doesn't fail on armhf in not-reproducible buildd: > https://buildd.debian.org/status/package.php?p=flint Exactly, so IMHO, this seems like a toolchain bug for which the fix is in unstable but it has not propagated to testing yet. In either case, a completely different bug, apparently, than the one it was reported here. Thanks.
Bug#818990: Is it really a flint problem?
Hi, On 21/05/2016 01:46, Santiago Vila wrote: It still fails on reproducible builds on the "armhf" architecture, but only in testing, and the way it fails is now completely different than the way it failed on amd64 in the initial report. But it doesn't fail on armhf in not-reproducible buildd: https://buildd.debian.org/status/package.php?p=flint Snark on #debian-science
Bug#818990: Is it really a flint problem?
It still fails on reproducible builds on the "armhf" architecture, but only in testing, and the way it fails is now completely different than the way it failed on amd64 in the initial report.
Bug#818990: Is it really a flint problem?
Hi,
On 19/04/2016 12:22, Mattia Rizzolo wrote:
On Tue, Apr 19, 2016 at 10:04:31AM +0200, Julien Puydt wrote:
Hi,
(1) according to:
https://buildd.debian.org/status/package.php?p=flint=sid
the version 2.5.2-3 of flint built perfectly on amd64.
yeah, that's a past tense.
(2) more recently (yesterday) :
https://buildd.debian.org/status/package.php?p=flint=experimental
the version 2.5.2-4 of flint built perfectly on amd64.
that's a different version.
(3) many tests on different amd64 systems (three by myself, and a few others
from people on #debian-mentors) didn't have any issue with 2.5.2-4 either.
that's still a different version.
Don't try to mix versions thinking it's always the same.
This suggest the package isn't at fault and the rare systems where the
failures occurred should be double-checked for the real root of the problem.
Indeed, with the very same chroot, I can build -4, but -3 FTBFS.
Evidently, this:
-\left|\frac{a_{n-2}}{a_n}\right|^{\frac{1}{2}}, \dots
+\left|\frac{a_{n-2}}{a_n}\right|^{\frac{1}{2}}, \dotsc
Is enough to fix this bug, and it also makes sense, given that the
failure log we pointed to you fails while doing latex stuff.
I'd say: mark this bug as fixed in that version, add the number to the
changelog and upload to unstable (why in first case that one went to
experimental?)
Mattia, the original #818990 was a FTBFS running the unit tests of the
2.5.2-3 package. When investigating this FTBFS, I found another failure,
within the documentation. I added a patch fixing this documentation
issue to get 2.5.2-4 (will be in next upstream too since I forwarded, of
course). That is due to some changes in texlive, which made a typo turn
into an error. And it is for this 2.5.2-4 that Chris still complained,
and that I counted you among the two reporters of FTBFS.
So now the situation is :
- Chris finds the package faulty within the reproducible framework ;
- every other compilation test of 2.5.2-4 was a success.
Perhaps this bug report should be re-assigned to another package?
Snark on #debian-science
Bug#818990: Is it really a flint problem?
On Tue, Apr 19, 2016 at 10:04:31AM +0200, Julien Puydt wrote:
> Hi,
>
> (1) according to:
> https://buildd.debian.org/status/package.php?p=flint=sid
> the version 2.5.2-3 of flint built perfectly on amd64.
yeah, that's a past tense.
> (2) more recently (yesterday) :
> https://buildd.debian.org/status/package.php?p=flint=experimental
> the version 2.5.2-4 of flint built perfectly on amd64.
that's a different version.
> (3) many tests on different amd64 systems (three by myself, and a few others
> from people on #debian-mentors) didn't have any issue with 2.5.2-4 either.
that's still a different version.
Don't try to mix versions thinking it's always the same.
> This suggest the package isn't at fault and the rare systems where the
> failures occurred should be double-checked for the real root of the problem.
Indeed, with the very same chroot, I can build -4, but -3 FTBFS.
Evidently, this:
-\left|\frac{a_{n-2}}{a_n}\right|^{\frac{1}{2}}, \dots
+\left|\frac{a_{n-2}}{a_n}\right|^{\frac{1}{2}}, \dotsc
Is enough to fix this bug, and it also makes sense, given that the
failure log we pointed to you fails while doing latex stuff.
I'd say: mark this bug as fixed in that version, add the number to the
changelog and upload to unstable (why in first case that one went to
experimental?)
--
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
Bug#818990: Is it really a flint problem?
Hi, (1) according to: https://buildd.debian.org/status/package.php?p=flint=sid the version 2.5.2-3 of flint built perfectly on amd64. (2) more recently (yesterday) : https://buildd.debian.org/status/package.php?p=flint=experimental the version 2.5.2-4 of flint built perfectly on amd64. (3) many tests on different amd64 systems (three by myself, and a few others from people on #debian-mentors) didn't have any issue with 2.5.2-4 either. This suggest the package isn't at fault and the rare systems where the failures occurred should be double-checked for the real root of the problem. Thanks, Snark on #debian-science
