fix two report bugs caused by previous commit
ClosedPublic

Authored by adamwill on Dec 20 2015, 11:27 PM.

Details

Summary

Wow, I suck at coding and you guys suck at review ;). First:
the imagetype/payload discovery is now run on all jobs, but it
will crash for jobs where the flavor is 'universal', because
that doesn't yield the expected two values. So deal with that.
Second, it turns out we do need to deepcopy the testcase dict,
because we get each dict multiple times; when we modify the
dict in place we poison it for future runs - e.g. if the first
job we hit for some testcase happens to be the BIOS job, the
dict's env gets set to 'x86 BIOS', and when we hit the job for
the same testcase but for UEFI, the env remains 'x86 BIOS'
because the magic value has already been replaced in the dict.
So we need to copy the dict each time we read it out of
TESTCASES.

Note: this is why the 2015-12-19 results did not initially
show up in the wiki.

Test Plan

Run the result submission stuff for a full set of
tests. As long as some have 'universal' as the flavor, you
should see the crash and this should fix it. Note that for
tests where we cover more than one env - e.g. custom_btrfs -
the old code would only produce a ResTuple for one of the
envs (whichever job it hit first); the fixed code produces
ResTuples for each env.

Diff Detail

Repository
rOPENQA fedora_openqa
Lint
Automatic diff as part of commit; lint not applicable.
Unit
Automatic diff as part of commit; unit tests not applicable.
adamwill retitled this revision from to fix two report bugs caused by previous commit.Dec 20 2015, 11:27 PM
adamwill updated this object.
adamwill edited the test plan for this revision. (Show Details)
adamwill added reviewers: jskladan, garretraziel.
This revision was automatically updated to reflect the committed changes.

I've landed this now since I didn't want it to be broken for the whole shutdown. Please do note any problems with it, though.