Apache Allura / Chat is hosted on FreeNode IRC channel #allura
ImportError: No module named allura.tests.decorators
Um, I think you'll find that there quite is such a module, my friend.
However, allura is failing becuase testfixtures wasn't added to our local mirror
And yet Dave claims otherwise. Hrm
i added it last night
i think those will both fix themselves
there is a bug in in the projects git branch view if there's an / in the branch name
Is that known? Do we have a ticket for that? --^
i'm reviewing the fix now
that ticket is an excellent example of where displaying comments and metadata changes need improving, by the way
wjp: can you elaborate?
scanning through the comments/history of that ticket is very inefficient
because metadata changes take up more vertical space than is really justified and there is only a limited number of comments on a single page
good point about the metadata waste
yep, definitely. I've even poked around at the CSS a little bit trying to do a quick fix, but didn't have the time/skills to make a fast fix
and i agree about the paged comments too. seems like we might have a ticket for that already but not sure
do we know why that's failinlg?
b/c it can't import allura for some reason?
ah I figured out the forgehg problem
it tried to pull down Allura from pypi
because the Allura build (upstream job) had failed
so it wasnt' available
yeah, and allura is failing b/c it can't install testfixtures
which doesn't make sense - it was working last night
Allura doesn't trigger the forgehg (downstream) if it fails, but there were new forgehg commits which made it run anyway
i kicked off a new Allura run, because I agree it seems like it should work
wonder if we could add a rule to not build forgehg if allura is red
although i guess that would usually be a little extreme
since it will still usually be importable, even if it's failing
that looks like just a regular legit test failure
cory_fu: perhaps related to empty repo changes
hrm, i fixed that on my sandbox during testing
i'll take a look
i'm going to delete that test
it's hitting a url for a tool that doesn't exist, and not even checking the response
it's failing now because we hit the disk if the metadata is missing, and there's no svn repo there
Yippie, build fixed!
not sure what that test failure is about, lots of 404s in the functional tests
wtf, i ran all these tests repeatedly on my sandbox
and they still all pass for me on master
oh i know why
forgehg master is ahead of allura master, on which it depends
but i can't push allura master yet, b/c i'm waiting for the new forgehg lib to be installed in pythontree
Getting a couple recent reports of git browser not showing up-to-date info. I thought that was fixed. :(
Yeah, so, new ticket then?
Or roll into that one?
seems about the same as 6101 to me
It is git vs. hg.
cory_fu: you think that'd make it a different bug?
It showing a commit as modifying a directory that it didn't modify is quite possibly a separate bug from it showing a correct but out-dated set of commits
I'll make a new ticket and say "may be related to:" ;)
Yeah, I think that's best
I'm really sad that these types of issues are still cropping up
User notes that it seems only the root that's affected.
ctsai-sf: here, you mean? :P
Yes, this is indeed what I meant when I said that we were talking about your problem in #allura
Quibus is part of the openmsx project.
If you click the doc dir, you can see only recent stuff
while the message, user and date of the doc dir on root level is about 12 years old
(just as an example)
maybe the new 6152 should be closed in favor of that one
those are older commits, we probably need to run the last commit updating script on it
we haven't run that much at all yet
so keeping them separate tickets is probably ok
But these new tickets would presumably be using the newer LCD records so maybe there's still an issue with the new version
We had just had this repo pushed last night
Is that related?
Yes, it shows that this issue is not entirely the same as similar issues we've seen in the past. Hence the decision to keep separate tickets for this rather than closing one in favor of the other.
You could probably test by reproducing it by uploading it again on your test env.
Yippie, build fixed!