I submitted a patch that included a new .test file tonight (#874364-5: Create a file example demonstrating key file APIs and the streamwrapper concept) and it got tested (successfully): http://qa.drupal.org/pifr/test/78049

But if you look at that test, you'll see that the actual tests included in the patch (File Example tests) did *not* get run.

What I did: Submitted a patch that contained a new module with a new .test file.

What I expected: The tests in the .test file would be run

What happened instead: No tests from the patch were run.

Support from Acquia helps fund testing for Drupal Acquia logo

Comments

boombatower’s picture

Project: Project Issue File Review » Project issue file test

All part of #102102: Parse project .info files: present module list and dependency information and the way PIFT send instructions on what to test. Should use directory argument instead of module.

rfay’s picture

Oddly, the file_example tests *still* aren't being tested, even now that the patch is committed. Hints on this? Is there a workaround? I've even done a retest, and file_example never gets tested.

boombatower’s picture

Once your dev tarball is regenerated then it should (crappy limitation that will be removed).

rfay’s picture

I had to make a new commit before it actually got tested, but that did the job.

Oddly, see http://qa.drupal.org/pifr/test/26934, debug() messages are being shown on qa.drupal.org. I thought they would not be displayed...

boombatower’s picture

Nope...that was the original spec we decided on so people would know if they had them in their tests and could use them. Not sure we want to keep it that way but...

I am also going to guess that your dev tarball was rebuilt..because making another commit shouldn't do anything.

rfay’s picture

This is an insidious and horrible #fail.

boombatower’s picture

Title: New tests in a new .test file in a patch don't get executed » Use test.directory.review instead of module list

Using 'test.directory.review' argument instead of module list should solve the latency issue.

boombatower’s picture

Assigned: Unassigned » boombatower
Status: Active » Needs review
FileSize
907 bytes
jthorson’s picture

Re-rolled, with a couple tweaks to bring it up to date.

rfay’s picture

I'd say that if this is the way to do it, let's get it in. Let's just make sure that we keep the arguments lists up-to-date in http://drupal.org/automated-testing, as they're a lot of magic that gets made up on the fly.

+++ b/pift.cron.inc
@@ -328,7 +328,10 @@ function pift_cron_queue_batch_build_branches_process(array &$batch, array &$bra
+      $item['plugin_argument']['test.directory.review'][] = 'sites/default/modules/' . $project->project['uri'];

I do hate PIFT having knowledge of the directory structure being used on PIFR, but if this is the way to do it, I'm good with it.

jthorson’s picture

I do hate PIFT having knowledge of the directory structure being used on PIFR

That's an interesting point ... I'd suggest we put this in, but create an issue (for tracking/history purposes) for decoupling the directory structures (perhaps make these settable, and require the client and server to be set the same?).

In any case, the 'sites/default/modules' piece is actually referring to the temporary directory that is created inside the 'sites/default/files/checkout/' root only for the duration of the review; so we're not really exposing much.

rfay’s picture

It will be nice to get this in. I just did the same thing again (new test) and of course it doesn't get run.

boombatower’s picture

The current design of pifr/pift was in migration when I was told to stop...because of that remnants like this argument exist. The PIFT side should tell PIFR what to do for a long list of reasons so the fact that it doesn't tell PIFR where to checkout modules is simply something I planned down the road and did with RD.

jthorson’s picture

Committed to 6.x-2.x (9a9c30f)

jthorson’s picture

Status: Needs review » Fixed

Status: Fixed » Closed (fixed)

Automatically closed -- issue fixed for 2 weeks with no activity.