Current upload for 10.1.28-1 fails dep8 unitest on s390x

[Christian Ehrhardt ]

Hi,
this is all kind of interwoven with each other, so I try to summarize the bit I thought to understand.

1. the last upload of August worked and still does on all architectures [1]
2. there were issues around pcre3 related tests [2]
   Those were fixed upstream [3], but partially as changes to the local pcre3
3. The same change was done to the pcre3 lib of Artful in [4]
   (be aware those pcre versiosn are not fully the same)
4. But our unit tests are still failing (on s390x).
   In fact the tests that now fail on the .28 [5] upload are new.
   That way it is not actually a regression.
5. I tried to adapt the test changing bits of [3] to [5] but it still fails

Since this is s390x only and the test is new rather than a regression I'd suggest to unblock for now by adding a arch dependent skip for now.

[1]: https://launchpad.net/ubuntu/+source/mariadb-10.1/10.1.25-1
[2]: https://jira.mariadb.org/browse/MDEV-13412
[3]: https://github.com/MariaDB/server/commit/6454d4e17727b6c786fbb953f4cc05c8cd739e83
[4]: https://launchpad.net/ubuntu/+source/pcre3/2:8.39-5ubuntu1
[5]: https://launchpad.net/ubuntu/+source/mariadb-10.1/10.1.28-1

[Christian Ehrhardt  (paelzer)]

Testing my suggestion at the moment and will then upload a git entry to review and RFC from people related to these changes/packages.

[Christian Ehrhardt  (paelzer)]

The following change would be the "unblocker" for now.
I lack any better idea as after some time I'm lost in the rabbit hole of what actually goes on in this test between pcre and mariadb - I didn't reach the stack size detection that was broken last time.

So as outlined initially this will unblock the current upload without loosing all of the dep8 coverage.

=> https://git.launchpad.net/~paelzer/ubuntu/+source/mariadb-10.1/commit/?h=unblock-s390x-dep8&id=a500d2177016c281d6c6999b9217df5e5b0e5dc7

[Christian Ehrhardt  (paelzer)]

Subscribing Robie and Ondřej to get their feedback as they worked on the last related uploads.

[Robie Basak (racb)]

+1 to skip this test, IMHO. Ondřej, can you sort this in Debian please? I'd like to avoid an Ubuntu delta.

[Ondřej Surý (ondrej)]

Looks good to me.

[Robie Basak (racb)]

mariadb-10.1 (1:10.1.29-2) unstable; urgency=medium

  [ Otto Kekäläinen ]
  * Update the d/changelog with CVEs

  [ Ondřej Surý ]
  * Revert to using system pcre library (Closes: #882329)
  * Bump the epoch to fix the mess created by mariadb-10.2 upload
    (Closes: #881898)

  [ Christian Ehrhardt ]
  * d/t/upstream: skip func_regexp_pcre on s390x

 -- Ondřej Surý <email address hidden> Wed, 22 Nov 2017 06:03:17 +0000

But 1:10.1.29-2 is currently stuck in proposed due to Debian bug 888956