Fix load_templates --clean to wipe more than one entry per table #2955
Conversation
AdamWill
force-pushed
the
fix-templates-clean
branch
3 times, most recently
from
7d054ab to
edbef94
Compare
|
The failure here doesn't look like anything to do with the PR. |
|
yes, ui test failure is unrelated, should be fixed meanwhile with a PR merged to master, retriggered test here. |
There was a problem hiding this comment.
This makes generally sense. But let's have @kalikiana have a look at it, too.
| @@ -241,7 +241,7 @@ if ($options{'clean'}) { | |||
| my $id = $result->{$table}->[$i]->{id}; | |||
| my $table_url_id = $url->clone->path($options{'apibase'} . '/' . decamelize($table) . "/$id"); | |||
| $res = $client->delete($table_url_id)->res; | |||
| last if $res->is_success; | |||
| last unless $res->is_success; | |||
There was a problem hiding this comment.
I did not realize I'd flipped this condition. Good catch!
| test_once $args, $expected, 'imported original fixtures'; | ||
| is $schema->resultset('Machines')->count, 1, "only one machine is loaded"; | ||
| my $machine = $schema->resultset('Machines')->first; | ||
| is $machine->name, "32bit", "correct machine is loaded"; |
There was a problem hiding this comment.
That looks great. Thank you for coming up with the test.
1400d8a inverted the logic here - instead of exiting the loop on a *failure* we now exit it on *success*. So each time --clean is run it will only wipe one entry per table. This obviously breaks things badly (the clean phase doesn't clean things right at all, so the subsequent load phase fails because we try to add entries that already exist, and you wind up with a mess). This flips the logic back, and adds a test. Signed-off-by: Adam Williamson <awilliam@redhat.com>
okurz
force-pushed
the
fix-templates-clean
branch
from
edbef94 to
5396d7f
Compare
|
I updated your branch. That should fix the failed test. |
Codecov Report
@@ Coverage Diff @@
## master #2955 +/- ##
==========================================
+ Coverage 93.44% 93.46% +0.01%
==========================================
Files 190 190
Lines 11959 11959
==========================================
+ Hits 11175 11177 +2
+ Misses 784 782 -2
Continue to review full report at Codecov.
|
1400d8a inverted the logic here - instead of exiting the loop
on a failure we now exit it on success. So each time --clean
is run it will only wipe one entry per table. This obviously
breaks things badly (the clean phase doesn't clean things right
at all, so the subsequent load phase fails because we try to add
entries that already exist, and you wind up with a mess).
This flips the logic back, and adds a test.
Signed-off-by: Adam Williamson awilliam@redhat.com