test results in php5-common are more than 1 MiB larger than last upload

Bug #583933 reported by Micah Gersten
8
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Ubuntu Server papercuts
Invalid
Low
Clint Byrum
php5 (Ubuntu)
Confirmed
Low
Unassigned

Bug Description

Binary package hint: php5

I noticed that usr/share/doc/php5-common/test-results.txt.gz was a little over a 1 MiB larger in the latest upload for php5-common:

https://edge.launchpad.net/ubuntu/+source/php5/5.3.2-1ubuntu4.1/+build/1715178
php5-common_5.3.2-1ubuntu4.1_amd64.deb (533.8 KiB)

https://edge.launchpad.net/ubuntu/+source/php5/5.3.2-1ubuntu4.2/+build/1738478
php5-common_5.3.2-1ubuntu4.2_amd64.deb (1.6 MiB)

-rw-r--r-- 1 micah micah 228382 May 4 02:08 php5_4.1/usr/share/doc/php5-common/test-results.txt.gz
-rw-r--r-- 1 micah micah 1455102 May 13 15:20 php5_4.2/usr/share/doc/php5-common/test-results.txt.gz

Chuck Short (zulcss)
Changed in php5 (Ubuntu):
importance: Undecided → Low
status: New → Confirmed
Thierry Carrez (ttx)
Changed in server-papercuts:
importance: Undecided → Low
milestone: none → maverick-alpha-2
status: New → Triaged
Changed in php5 (Ubuntu):
assignee: nobody → Clint Byrum (clint-fewbar)
status: Confirmed → In Progress
Revision history for this message
Clint Byrum (clint-fewbar) wrote :

It appears that the tests failed for some reason in 5.3.2-1ubuntu4.2, but were successful again in 5.3.2-1ubuntu5, as the file is the right size.

However, the bug is not fixed, as the failed tests should have failed the build.

I believe it has to do with piping the output of the tests to tee, therefore missing the exit code and not failing the tests... patch forthcoming.

Thierry Carrez (ttx)
Changed in server-papercuts:
assignee: nobody → Clint Byrum (clint-fewbar)
Revision history for this message
Clint Byrum (clint-fewbar) wrote :

Upon closer inspection, it would appear that there are always a few tests that fail. I will be adding a check to make sure the number of failed tests does not exceed half the number of passed tests, which should be lenient enough for the loose nature of the PHP tests, but detect seriously broken tests and/or builds.

Revision history for this message
Clint Byrum (clint-fewbar) wrote :

Added a clause to debian/rules that checks test-results.txt to see if the number of FAIL's are more than half of the number of PASS's. This will prevent not only a giant test-results.txt, but also a severely broken .deb from being built.

Changed in php5 (Ubuntu):
status: In Progress → Confirmed
assignee: Clint Byrum (clint-fewbar) → nobody
Revision history for this message
Chuck Short (zulcss) wrote :

Clint,

I dont think this is the correct way to go about it. What we should probably do is identify which tests are actually good compared to which tests are bad and then basically do a divide and conquer. Ideally I would like us to fix the flakey tests as well.

chuck

Revision history for this message
Clint Byrum (clint-fewbar) wrote :

After spending a few hours on this, I think this does in fact need more attention. However, its going to require even more time, so I'd propose we remove it from server papercuts.

Thierry Carrez (ttx)
Changed in server-papercuts:
status: Triaged → Invalid
Revision history for this message
Chuck Short (zulcss) wrote :

Nothin to sponsor here.

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.