When I submit a patch to an issue, then the test bot will ignore it if
- I forget to set it to "needs review"
- the patch does not have a patchy name
- there was more than one file uploaded, and the test bot will only test one of them.
For someone who is using this system for the first time, or has not used it for a few months, or who is just a bit tired, this system is totally intransparent and mysterious. As a result, we see series of comments with patches, all trying to activate the test bot. Here is one issue where I was the lucky one.
This does not just bloat the issue, it also makes the person (me, in this case) look like an idiot. We really should try to avoid that.
So, as a first and simple (I hope) solution, we should change or enhance the generic "Ignored" message, and specify why the patch was ignored, or why it is not considered a patch.
Or at least, display a link next to the comment form, to a page where this stuff is documented.
Comment | File | Size | Author |
---|---|---|---|
#18 | pift.specify_ignored_if_due_to_status-1162082-18.patch | 3.32 KB | jthorson |
#14 | pift.specify_ignored_if_due_to_status-1162082-14.patch | 3.35 KB | jthorson |
Comments
Comment #1
donquixote CreditAttribution: donquixote commentedAs an advanced solution, there could be a widget telling you which of the uploads will be tested, before you submit the comment form.
Comment #2
rfayI certainly agree with this feature request! There are so many tweaky things about this, and this is one we can deal with fairly easily.
Another reason a test can be ignored: Has a name like -d6.patch. I'm going to make that name more descriptive though and should solve that one.
Great ideas.
Comment #3
donquixote CreditAttribution: donquixote commentedAnother example,
#512962-244: Optimize menu_router_build() / _menu_router_save()
Patch name was "drupal.menu_router_build.SELECT.d8.patch"
Status was "needs review".
Another test was still running (no idea if that matters).
Comment #4
donquixote CreditAttribution: donquixote commentedComment #5
rfayHere, of course, it's the .d8.patch that causes it to be ignored. That's the (weird) formula for ignoring patches. We have to change it to something more obvious.
Comment #6
donquixote CreditAttribution: donquixote commentedIs this formula documented somewhere?
And what is the point? For me, .d8.patch sounds like the most legitimate name for a D8 patch ever.
Comment #7
rfayIt *was*. Can't find it. Totally bizarre IMO.
Comment #8
rfayMoving to PIFT, where it belongs.
Comment #9
Anonymous (not verified) CreditAttribution: Anonymous commentedI tried running a patch and it was ignored. I look at a simular patch and examined the patch documentation. I follow the procedures for the layout including where about it was located using the @@. This is my first patch so i am not sure why it was ignored.
Comment #10
rfay@alexdrupaldev when you have a problem like that give us a link to the problem patch so we can explore it.
I also find it really annoying when it just says "ignored". Today I was debugging a problem which was caused by no test being inserted into the database. Same stuff. "Ignored".
Comment #11
kathyh CreditAttribution: kathyh commentedAgreed. Would be nice to have a comment on WHY it's ignored AND let the user EDIT the previous post to switch it to "needs review". I can edit the comment but not the settings, so have to add additional posts which leads to the bloat and are uneccessary.
Comment #12
drupal_was_my_past CreditAttribution: drupal_was_my_past commented@rfray I also ran into the same problem as @alexdrupaldev. Also my first patch contribution to core: #806992: Run-on sentence in authorize.inc comment #13. Any ideas how to properly create a patch for core? Should I be running git diff with the --no-prefix flag?
Comment #13
drupal_was_my_past CreditAttribution: drupal_was_my_past commentedOoops. I failed to set the correct status flag. Please ignore comment 12.
Comment #14
jthorson CreditAttribution: jthorson commentedHere's a first step ... should identify when issues are ignored due to status, and say so in the 'ignore' message.
Comment #15
rfayHow about saying even more helpful "Only issues in states 'needs review' or 'rtbc' get tested" ?
Also note that variable pift_test defaults to array(8, 14), not an empty array.
But I like this as a first step.
Can we rejigger the "ignore on weird filename" behavior in this patch, and make sure to give feedback on it?
Comment #16
jthorson CreditAttribution: jthorson commentedBoth the pift_status array and labels used for issue statuses are drupal.org dependent, and may not apply on a non-d.o site. While I admit that d.o is the prime use case, we shouldn't hard-code any assumptions.
How about simply 'Ignored: Check issue status.'?
Comment #17
rfayI'm ok with that but the default is in fact already wired in by the install process even though it's not in the code. The variable empty case never happens because the var is set in the install if I'm not mistaken. But it's ok. We have a ways to go with this one anyway.
Comment #18
jthorson CreditAttribution: jthorson commentedReworded the 'Ignored' message.
This was a 'low-hanging fruit' fix, in that the information required to check the issue status was already available in the functions which call the 'theme' function in which the ignored message is set. This isn't really the case for the other potential 'ignored' reasons.
There is a ways to go on this particular issue ... but I don't see any reason why we would need to wait until we've fixed 'all' ignored states before committing a quick hitter like this in the meantime.
Comment #19
jthorson CreditAttribution: jthorson commentedComment #20
rfayI didn't test this on the test environment (we could fiddle with pift_test to do that - you probably already have) but it looks good to me.
Unless.. you think we should just refactor to come up with a comprehensive approach to this.
Comment #21
jthorson CreditAttribution: jthorson commentedI think we should commit this, but leave the issue open for a more comprehensive refactoring.
Comment #22
jthorson CreditAttribution: jthorson commentedCommitted the 'Ignored: check issue status' update in #18 to 6.x-2.x (3735e61).
Setting issue back to 'active' for followup on other ignore reasons.
Comment #23
alexharries CreditAttribution: alexharries commentedFollowing this topic - such an annoying issue!
Comment #25
drummWith #1171958: Allow files to be assigned to branch(es)/version(s) and thus tested against it, there is now a check in JS for a diff/patch file being selected. This would be a good place to put in a warning if the issue status isn't in a testable state.