Bug 210589
| Summary: | databases/py-sqlalchemy-migrate: Update to 0.10.0 | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Product: | Ports & Packages | Reporter: | Kurt Jaeger <pi> | ||||||
| Component: | Individual Port(s) | Assignee: | Kurt Jaeger <pi> | ||||||
| Status: | Closed FIXED | ||||||||
| Severity: | Affects Some People | CC: | lwhsu, python | ||||||
| Priority: | --- | Keywords: | needs-qa, patch, regression | ||||||
| Version: | Latest | Flags: | pi:
maintainer-feedback+
koobs: maintainer-feedback+ |
||||||
| Hardware: | Any | ||||||||
| OS: | Any | ||||||||
| Attachments: |
|
||||||||
Created attachment 171823 [details]
patch-v2
this builds fine in poudriere on 10.3a+i, 9.3a. It builds on 11 outside of poudriere. Why does libreoffice take so long to build on 11 ? 9 hours ? This blocks my testjail for 11a 8-}
Thanks for the patch, and sorry that I cannot verify this very soon (in two weeks). You're welcomed to commit the update you think it's right, and please feel free to take over the maintainership. Committed, thanks. A commit references this bug: Author: pi Date: Wed Aug 3 16:21:44 UTC 2016 New revision: 419532 URL: https://svnweb.freebsd.org/changeset/ports/419532 Log: databases/py-sqlalchemy-migrate: 0.7.2 -> 0.10.0 PR: 210589 Changes: https://github.com/openstack/sqlalchemy-migrate/releases Approved by: lwhsu Changes: head/databases/py-sqlalchemy-migrate/Makefile head/databases/py-sqlalchemy-migrate/distinfo head/databases/py-sqlalchemy-migrate/pkg-descr Re-opening due to regression post commit: r419532 [1] caused ports that depend on sqlalchemy07 and sqlalchemy-migrate (at least buildbot) to fail because py-sqlalchemy-migrate pulls in py-sqlalchemy10 which conflicts with py-sqlalchemy07. http://beefy10.nyi.freebsd.org/data/110i386-default/419570/logs/buildbot-0.8.12_1.log [1] http://svnweb.freebsd.org/changeset/ports/419532 There needs to be 07 and 10 versions of -migrate unless all ports depending on 07 can be upgraded to use 10. More precisely there needs to be an equal number of *-migrate ports as there are sqlalchemy ports supported at all times, unless the latter can, and have been confirmed and tested to work with multiple versions of *-migrate. Can you revert this commit until a proper root-cause / fix is identified please If you change the dependency from sqlalchemy07 to sqlalchemy10, does buildbot still work ? btw, testbuilds for buildbot with sqlalchemy10 as dependency are fine. The only tests I would consider satisfying would be run time test suites (which buildbot does have). Having said that, there's 15 ports that currently RUN_DEPEND on sqlalchemy07, and switching their dependencies forward and testing properly will take longer than we want to have these ports remain broken. In the meantime we need a resolution quickly, and since reverting the change will also mean bumping PORTEPOCH (which lwhsu doesn't like), perhaps an alternative is creation of a py-sqlalchemy07-migrate and moving the current dependents over to that. Either way, we need to resolve this asap, either with an alternative, or failing one being viable, a revert and PORTEPOCH++ Ran 3032 tests in 17.243s FAILED (skips=27, errors=2, successes=3003) In the normal case, do all tests work ? (In reply to Kurt Jaeger from comment #9) A test with the previous setup: Ran 3032 tests in 37.556s FAILED (skips=27, errors=2, successes=3003) so updating buildbot to use py-sqlalchemy10 does not change the outcome of the tests ? (In reply to Kurt Jaeger from comment #10) Seems so, updating buildbot's dependency doesn't introduce more regressions. And there are just few ports depend on py-sqlalchemy-migrate: $ grep sqlalchemy-migrate /usr/ports/INDEX-10 | awk -F'|' '{ print $2}' /usr/ports/databases/py-sqlalchemy-migrate /usr/ports/devel/buildbot /usr/ports/multimedia/py-openlp /usr/ports/www/py-imdbpy Is it possible for you to check others? If the results are the same, we can update dependency of these ports. Yes, I'm working to find fixes for the others. I'm trying to avoid the revert... (In reply to Kurt Jaeger from comment #12) Thanks for the help, if that takes some more time, we can just add py-sqlalchemy-migrate07 in the mean time. That makes ports tree consistent. A commit references this bug: Author: pi Date: Sun Aug 7 10:07:53 UTC 2016 New revision: 419767 URL: https://svnweb.freebsd.org/changeset/ports/419767 Log: www/py-imdbpy: chase sqlalchemy dependencies PR: 210589 Changes: head/www/py-imdbpy/Makefile @Kurt I'll switch buildbot over to sqlalchemy10 shortly A commit references this bug: Author: koobs Date: Sun Aug 7 14:06:57 UTC 2016 New revision: 419777 URL: https://svnweb.freebsd.org/changeset/ports/419777 Log: devel/buildbot: Update sqlalchemy RUN_DEPENDS r419532 [1] caused ports that depend on sqlalchemy07 and sqlalchemy-migrate (at least buildbot) to fail because py-sqlalchemy-migrate pulls in py-sqlalchemy10 which conflicts with py-sqlalchemy07. Update the sqlalchemy dependency from 0.7.x to 1.0.x after QA testing for regressions @ runtime using the buildbot test suite. [1] http://svnweb.freebsd.org/changeset/ports/419532 PR: 210589 Reported by: pkg-fallout Tested by: pi MFH: 2016Q3 Changes: head/devel/buildbot/Makefile A commit references this bug: Author: koobs Date: Sun Aug 7 14:10:19 UTC 2016 New revision: 419778 URL: https://svnweb.freebsd.org/changeset/ports/419778 Log: MFH: r419777 devel/buildbot: Update sqlalchemy RUN_DEPENDS r419532 [1] caused ports that depend on sqlalchemy07 and sqlalchemy-migrate (at least buildbot) to fail because py-sqlalchemy-migrate pulls in py-sqlalchemy10 which conflicts with py-sqlalchemy07. Update the sqlalchemy dependency from 0.7.x to 1.0.x after QA testing for regressions @ runtime using the buildbot test suite. [1] http://svnweb.freebsd.org/changeset/ports/419532 PR: 210589 Reported by: pkg-fallout Tested by: pi Approved by: ports-secteam (blanket) Changes: _U branches/2016Q3/ branches/2016Q3/devel/buildbot/Makefile (In reply to Li-Wen Hsu from comment #11) multimedia/py-openlp has been fixed by the maintainer So, any ports left that need fixing ? I'll close this PR in 24h, if no other ports come up. I'll wait for the next pkg-fallout run, unless there's somewhere public? to see if the last run was successful for the previously failing ports devel/buildbot fails in quarterly, because the update was not MFHed. TODO: think about if a MFH would be useful -- I think then other updates would need MFH, which is probably excessive. Hm... Next quarterly is done, so probably all fine. If not, please create a new PR. |
Created attachment 171822 [details] patch portlint is fine. poudriere still needs some work