Bug#902557: transition: Perl 5.28
On 28/11/2018 06:29, Andreas Beckmann wrote: > Followup-For: Bug #902557 > > There is subversion/experimental still depending on the older perl. > > nmu subversion_1.11.0~rc2-1 . ANY . experimental . -m "Rebuild against > perlapi-5.28.0." Done. Emilio
Bug#902557: transition: Perl 5.28
Followup-For: Bug #902557 There is subversion/experimental still depending on the older perl. nmu subversion_1.11.0~rc2-1 . ANY . experimental . -m "Rebuild against perlapi-5.28.0." Andreas
Bug#902557: transition: Perl 5.28
On Wed, Oct 31, 2018 at 10:04:21AM +0100, Emilio Pozuelo Monfort wrote: > Control: tags -1 confirmed > Yes. Let's go ahead then as things look good now. Thanks, uploaded. -- Niko
Bug#902557: transition: Perl 5.28
Control: tags -1 confirmed On 28/10/2018 13:24, Mattia Rizzolo wrote: > On Sun, Oct 28, 2018 at 02:14:01PM +0200, Niko Tyni wrote: >> The only thing I found is gnumeric (#912099). Given it only links against >> libperl5.26, I'm not sure if that's an actual blocker or if libperl5.26 >> and libperl5.28 can exist in testing at the same time ("soft updates"?) > > If libperl can really be soft-updated these days, it doesn't block the > starting of the transition, but it *does* block the ending, as of course > we would like to remove libperl5.26 at some point... libperl amounts for a very small part of the transition, and that shouldn't be a problem. The larger issue is that all the perlapi-* rdeps need to migrate simultaneously, as they are provided by perl-base and thus can't both co-exist in testing. > Of course, gnumeric can be removed :P Yes. Let's go ahead then as things look good now. Cheers, Emilio
Bug#902557: transition: Perl 5.28
On Sun, Oct 28, 2018 at 02:14:01PM +0200, Niko Tyni wrote: > The only thing I found is gnumeric (#912099). Given it only links against > libperl5.26, I'm not sure if that's an actual blocker or if libperl5.26 > and libperl5.28 can exist in testing at the same time ("soft updates"?) If libperl can really be soft-updated these days, it doesn't block the starting of the transition, but it *does* block the ending, as of course we would like to remove libperl5.26 at some point... Of course, gnumeric can be removed :P -- 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#902557: transition: Perl 5.28
On Sat, Oct 27, 2018 at 03:07:18PM +0300, Niko Tyni wrote: > On Tue, Oct 02, 2018 at 09:21:58AM +0200, Emilio Pozuelo Monfort wrote: > > On 08/09/2018 13:37, Niko Tyni wrote: > > > > Release team: please consider granting us a transition slot once > > > the openssl situation has been resolved, and let me know if I've > > > missed something obvious. > > > > Please ping us again once that happens, in case we miss it. > > Hi, just noting that openssl 1.1.1 is in testing now (as I'm sure you know :) > > I've scheduled one more round of test binNMU rebuilds in case something > unrelated has regressed. Those will take a day or two to finish. > I'll report back when I have the results. The only thing I found is gnumeric (#912099). Given it only links against libperl5.26, I'm not sure if that's an actual blocker or if libperl5.26 and libperl5.28 can exist in testing at the same time ("soft updates"?) > I think libimage-sane-perl binNMU migration would get blocked by #908681 > (libsane1: pointless package rename), but there's just one rdep (gscan2pdf) > so testing removal seems like a viable option. This still applies. -- Niko
Bug#902557: transition: Perl 5.28
On Tue, Oct 02, 2018 at 09:21:58AM +0200, Emilio Pozuelo Monfort wrote: > On 08/09/2018 13:37, Niko Tyni wrote: > > Release team: please consider granting us a transition slot once > > the openssl situation has been resolved, and let me know if I've > > missed something obvious. > > Please ping us again once that happens, in case we miss it. Hi, just noting that openssl 1.1.1 is in testing now (as I'm sure you know :) I've scheduled one more round of test binNMU rebuilds in case something unrelated has regressed. Those will take a day or two to finish. I'll report back when I have the results. I think libimage-sane-perl binNMU migration would get blocked by #908681 (libsane1: pointless package rename), but there's just one rdep (gscan2pdf) so testing removal seems like a viable option. -- Niko
Bug#902557: transition: Perl 5.28
On 08/09/2018 13:37, Niko Tyni wrote: > Control: block -1 with 907015 > > On Mon, Sep 03, 2018 at 05:11:30AM +, Damyan Ivanov wrote: > >> * no packages that failed with perl 5.28 succeed with perl 5.26 > > Awesome. Thanks again for running these checks. > > The only blocker this transition that I'm aware of is openssl: binNMUs > of libnet-ssleay-perl for Perl 5.28 would not migrate to testing without > openssl, whose migration is blocked by #907015 . > > Release team: please consider granting us a transition slot once > the openssl situation has been resolved, and let me know if I've > missed something obvious. Please ping us again once that happens, in case we miss it. Thanks, Emilio
Bug#902557: transition: Perl 5.28
Control: block -1 with 907015 On Mon, Sep 03, 2018 at 05:11:30AM +, Damyan Ivanov wrote: > * no packages that failed with perl 5.28 succeed with perl 5.26 Awesome. Thanks again for running these checks. The only blocker this transition that I'm aware of is openssl: binNMUs of libnet-ssleay-perl for Perl 5.28 would not migrate to testing without openssl, whose migration is blocked by #907015 . Release team: please consider granting us a transition slot once the openssl situation has been resolved, and let me know if I've missed something obvious. Thanks for your work, -- Niko
Bug#902557: transition: Perl 5.28
-=| Niko Tyni, 15.08.2018 18:56:32 +0300 |=- > On Mon, Aug 13, 2018 at 04:58:47PM +, Damyan Ivanov wrote: > > -=| Niko Tyni, 12.08.2018 12:03:42 +0300 |=- > > > > > > I believe Damyan (cc'd) is now separately testing the whole archive > > > against Perl 5.28 on his own gear. Last I heard it should be finished > > > in about a week and the first 20% or so hadn't yielded any regressions, > > > but maybe he can give a status update? > > > > Current status: > > > > - 11 036 out of 28 414 source packages tried (39%), of which > > - 551 ftbfs with regular sid, as well as with perl 5.28 > > - 2 ftbfs with perl 5.28 and build fine with regular sid (https://people.debian.org/~dmn/perl-5.28) Status update: * 198 packages remain to be built * 1187 fail to build with perl 5.28 *and* with perl 5.26 (ftbfs/) * 3 failed with internal compiler errors and are to be retried (timeout/) * no packages that failed with perl 5.28 succeed with perl 5.26 * The failure of linbet-ssleay-perl with openssl 1.1.1~~pre9 is fixed in unstable (#895959). The rebuilding now proceeds with the packages that didn't finish rebuilding within 30 minutes (timeout.old/) and progress is understandably slow. The system shows somewhat unstable, with sporadic segmentation faults and internal compiler errors that disappear when the build is retried. -- dam
Bug#902557: transition: Perl 5.28
On Mon, Aug 13, 2018 at 04:58:47PM +, Damyan Ivanov wrote: > -=| Niko Tyni, 12.08.2018 12:03:42 +0300 |=- > > > > I believe Damyan (cc'd) is now separately testing the whole archive > > against Perl 5.28 on his own gear. Last I heard it should be finished > > in about a week and the first 20% or so hadn't yielded any regressions, > > but maybe he can give a status update? > > Current status: > > - 11 036 out of 28 414 source packages tried (39%), of which > - 551 ftbfs with regular sid, as well as with perl 5.28 > - 2 ftbfs with perl 5.28 and build fine with regular sid Thanks! > The two failures are strange, and I can't really attribute them to the > perl upgrade, but the fact remains - they build fine with sid and fail > with perl 5.28 installed. The python-stetl issue is #906222 in gdal. I don't think this should be a transition blocker, it's not Perl specific in any way. The bzr package builds fine for me against Perl 5.28, but seems to have sporadic build failures at https://tests.reproducible-builds.org/bzr . So probably transient. Release team: is there a long transition queue ahead of us? How soon would it be feasible to start the transition? Do you want to wait for Dam's rebuilds to finish before going ahead? Clearly there are going to be very few issues. -- Niko
Bug#902557: transition: Perl 5.28
-=| Niko Tyni, 12.08.2018 12:03:42 +0300 |=- > > We still want to do a full archive test rebuild soonish: as > > perl-base is > > essential and perl is transitively build essential, there are implicit > > build dependencies in the archive that we haven't covered yet. I don't > > expect much fallout though. I'll update this bug once we have more data. > > This has been delayed due to a shortage of disk space on perl.debian.net. > We've only been able to test rebuild the 3000 or so packages directly > depending on perl. > > I believe Damyan (cc'd) is now separately testing the whole archive > against Perl 5.28 on his own gear. Last I heard it should be finished > in about a week and the first 20% or so hadn't yielded any regressions, > but maybe he can give a status update? Current status: - 11 036 out of 28 414 source packages tried (39%), of which - 551 ftbfs with regular sid, as well as with perl 5.28 - 2 ftbfs with perl 5.28 and build fine with regular sid The two failures are strange, and I can't really attribute them to the perl upgrade, but the fact remains - they build fine with sid and fail with perl 5.28 installed. Perhaps this is related to the methodology. The 'perl 5.28' chroot has perl/5.28 installed into id, while the 'plain sid' one does not. Earlier there were failures related to usrmerge -- debootstrap now creates chroots with merged /usr. I have reverted that and used a clone of an old 'plain sid' chroot, which was created with an older debootstrap to avoid the noise. The logs of the two failed packages are on https://people.debian.org/~dmn/perl-5.28/failed/ (and I'll keep that updated as the rebuilds progress). -- dam
Bug#902557: transition: Perl 5.28
Control: block -1 with 905913 On Sun, Aug 12, 2018 at 12:03:42PM +0300, Niko Tyni wrote: > On Wed, Jun 27, 2018 at 10:27:08PM +0300, Niko Tyni wrote: > > Package: release.debian.org > > Severity: normal > > User: release.debian@packages.debian.org > > Usertags: transition > > Forwarded: https://release.debian.org/transitions/html/perl5.28.html > > > Perl 5.28 is in experimental and we'd like to get it into buster > > at some point, so filing this a bit pre-emptively. > > Hi, update on this: all the known blockers are now resolved. Oh, I see there's an uncoordinated (and apparently accidental) libsane transition that's going to block this. #905913 would prevent a Perl 5.28 rebuild of libimage-sane-perl from entering testing. -- Niko Tyni nt...@debian.org
Bug#902557: transition: Perl 5.28
On Wed, Jun 27, 2018 at 10:27:08PM +0300, Niko Tyni wrote: > Package: release.debian.org > Severity: normal > User: release.debian@packages.debian.org > Usertags: transition > Forwarded: https://release.debian.org/transitions/html/perl5.28.html > Perl 5.28 is in experimental and we'd like to get it into buster > at some point, so filing this a bit pre-emptively. Hi, update on this: all the known blockers are now resolved. In particular, qt4-perl is no longer required by debconf (thanks Colin!), and has been requested for removal in #904583. > We still want to do a full archive test rebuild soonish: as perl-base is > essential and perl is transitively build essential, there are implicit > build dependencies in the archive that we haven't covered yet. I don't > expect much fallout though. I'll update this bug once we have more data. This has been delayed due to a shortage of disk space on perl.debian.net. We've only been able to test rebuild the 3000 or so packages directly depending on perl. I believe Damyan (cc'd) is now separately testing the whole archive against Perl 5.28 on his own gear. Last I heard it should be finished in about a week and the first 20% or so hadn't yielded any regressions, but maybe he can give a status update? -- Niko
Bug#902557: transition: Perl 5.28
Package: release.debian.org Severity: normal User: release.debian@packages.debian.org Usertags: transition Forwarded: https://release.debian.org/transitions/html/perl5.28.html X-Debbugs-Cc: p...@packages.debian.org Control: block -1 with 899021 900832 901080 901082 901085 887687 862678 900232 902556 Dear release team, Perl 5.28 is in experimental and we'd like to get it into buster at some point, so filing this a bit pre-emptively. We've been doing test rebuilds of all reverse build dependencies of perl for a while on perl.debian.net, and things are looking pretty good, with the hairiest issue being #887687 (debconf needs qt4-perl which doesn't seem to have a bright future at all). I'm marking this bug as blocked by the handful of 5.28 related regressions we've spotted, and another handful of unrelated build failures in current sid that would prevent the necessary binNMUs. I've excluded packages not in testing, with the exception of libembperl-perl which Axel said he'll be looking at soon. We still want to do a full archive test rebuild soonish: as perl-base is essential and perl is transitively build essential, there are implicit build dependencies in the archive that we haven't covered yet. I don't expect much fallout though. I'll update this bug once we have more data. As usual, bugs are being filed with the usertag debian-p...@lists.debian.org / perl-5.28-transition . See https://bugs.debian.org/cgi-bin/pkgreport.cgi?tag=perl-5.28-transition;users=debian-p...@lists.debian.org Thanks for your work, -- Niko Tyni nt...@debian.org