Merge lp:~paelzer/britney/hints-ubuntu-focal-re2c into lp:~ubuntu-release/britney/hints-ubuntu

[Christian Ehrhardt ]

[Christian Ehrhardt  (paelzer)]

FYI: Debian doesn't run i386 tests of that https://ci.debian.net/packages/r/re2c/ (one less chance to compare)

[Steve Langasek (vorlon)]

> The first test since the change to i386-on-amd64 started failing with the
> new 1.3-1 upload that got synced from Debian.

The wording here can be read to imply that the tests started failing with the first run of tests after the switch to i386 on amd64. But this is not the case. The tests in the previous version of the package passed under i386-on-amd64 cross-testing, it's only the tests in the new version that are now failing. I don't think an all-version hint is appropriate here.

review: Disapprove

[Christian Ehrhardt  (paelzer)]

Thanks Steve for taking a look.

Yeah I meant tarting with 1.3.1 sorry to be misleading.

I might have been in too much all-version-i386 these days and got too zealous.
It might end up as a versioned override still, but ack on not all-version.

One of us will take another look sooner or later in our proposed-migration checks.

[lp:~paelzer/britney/hints-ubuntu-focal-re2c]

4464. By Christian Ehrhardt  on 2020-02-04

only restrict 1.3-1 of re2c

[Christian Ehrhardt  (paelzer)]

I have tried to test this on i386 myself, but without i386 images that isn't very easy anymore.
After being lost in dependency hell twice - and since Bryce said he tried and it worked (no test fail on local try) for him, I'm suggesting to override version 1.3-1 for now.

[Christian Ehrhardt  (paelzer)]

I have found that our infra uses -a i386 but even the latest autopkgtest git doesn't seem to know about that argument.

If there is a best practise for "how to run i386 tests like launchpad does" for the new i386-reality in focal I think I and the team would be interested. Let me know if such a thing exist.

[Christian Ehrhardt  (paelzer)]

Something else in the Archive or the autopkgtest-testers changed (feels bad to not know, but that is what happened).

The same 1.3-1 that formerly failed reliably now seems to test fine all the time, see http://autopkgtest.ubuntu.com/packages/r/re2c/focal/i386

Therefore I abandon this MP

review: Disapprove

[Christian Ehrhardt  (paelzer)]

I see the versioned hint got already merged.
With the recent results we can undo that again ?!?

I'll file another MP for it thou

=== modified file 'ubuntu-release'
--- ubuntu-release	2020-02-03 20:53:34 +0000
+++ ubuntu-release	2020-02-04 14:58:21 +0000
@@ -49,6 +49,8 @@
 force-badtest git-buildpackage/0.9.17 git-buildpackage/0.9.17ubuntu3
 ## x32 tests now run and fail on i386 with amd64 kernels, whee:
 force-badtest glibc/2.30-0ubuntu3/i386
+## i386 tests fail on i386 since 1.3-1 but are hard to reproduce outside of autopkgtest infra
+force-badtest re2c/1.3-1/i386
 ##
 force-badtest gvfs/1.42.2-1ubuntu3/amd64 gvfs/1.42.2-1ubuntu3/arm64
 ## the autopkgtests were not running the testsuite before but they do now