Apache Allura / Chat is hosted on FreeNode IRC channel #allura
Tim Van Steenburgh: Consider repo empty if it doesn't exist on disk.
does it make sense for a tool to assert what a valid mount point is for itself?
or should the platform always be in control of that...
one way is for Application to provide a default mount point pattern
and let subclasses override
but it seems backwards
otoh, i can't think of a better way to do it
another option could be for the "relaxed" mount point definition to be in the platform, and each Application can toggle that on
so each App doesn't have to have the full regex nor could an App go crazy and do something really weird
ah, i like that
i just realized that i inadvertently opened up project names also. gonna fix that now
ok, that's fixed
regarding escaping when going into solr and/or when coming out of solr (the text field is used to highlight matches in results)
They have different contents for comments.
i just saw the comment come in via email
I guess, if you comment on an old revision, it sticks with that old revision. But if you comment on the latest, it's going to stay with the latest.
every version of a wiki page is itself an Artifact
Yeah, I saw the email too, but I noticed that the link in the footer had the version.
we shouldn't allow that
subject line too
Currently default is "Developer" for CREATE.
brondsem: i'm more a fan of escaping on the way out
i like knowing that what's in the date store is the original content
and my fear is that we might not escape something properly on the way in
plus, i think you have to do it that way if you want highlighting to always work
and re-indexing everything to fix an issue like htat would take forever, whereas fixing the escaping on the way out is just a code push
like if you search for "this & that", highlighting won't work if you've escaped on the way in
i think i'll go that route then
ctsai-sf: agree on 3779, I would've thought it was like that already
So, I found where to change the default permissions, but, don't know how to fix the corresponding test.
which test is it?
and how does it fail with your change?
Just rolled back the fiddling I was doing with that test, and running tests again. Paste incoming shortly.
ha, that's a big list != to another big list
glad to see someone using the vagrant box :)
on the import line for assert_equal
change that to import from datadiff.tools instead of nose.tools
tvansteenburgh: Oh yeah, I meant to tell you I got that working over the weekend.
I probably should have asked this earlier. Can I run just that one test? Instead of all the ForgeTracker tests?
note the colon, the syntax is slightly different
if you install nose-progressive and use that, one of the nice features is that it formats your test results better so you can copy & paste the test names
to re-run them
ERROR: Failure: AttributeError ('module' object has no attribute 'test_ticket_model')
you in the ForgeTracker dir?
Oh, no. That would probably help.
and if you install nose-progressive, enable it with this in your ~/.noserc
yeah, i think datadiff only has some assert_* methods, so you'll have to still import 'raises' from nose.tools
tvansteenburgh: one issue with storing the verbatim text in solr is that it contains markdown formatting
so the reason we went down the sanitization path (when storing in solr) was to strip the markdown formatting and get plain text
maybe that can happen on the way out too....
hrm, didn't think of that
that's tricky b/c you don't really want to search against the markdwn
process is/was to convert markdown to html, and then strip the tags
it might be ok though, because most markdown is just special chars which solr will ignore
ctsai-sf: so it looks like that didn't help, because t.acl is a ming-instrumented list (hence the "I" prefix)
could try converting it to a normal list like list(t.acl) in the assert_equal line
Hrm, wasn't quite sure what you meant, but I tried using "import q; q(list(t.acl))" in a line just above, and it still has the "I" prefix.
i think you're going to have to compare the big lists by hand
t.acl._deinstrument() would give you the real "list", but datadiff's assert_equal throws an error on some of the contents of that list
I see. I'll ticket and poke at this more later. Have regular support stuff I need to work on for now.
ok. I think it'll end up being an easy change to the test. harder to figure out what the change needs to be :)
Is there a reference somewhere for tasks that can be run via /nf/admin/task_manager? (for PROSE docs)
find . -name '*.py' | xargs grep '@task' -A1
that is all the tasks defined in the code
and also: find . -name '*.py' | xargs grep ScriptTask
Hmm, I see. Possibly out of the scope for this doc then. I'll pass that on to them at least though.
well only the ScriptTask ones can be run from the web ui
Ah, that's helpful.
you can run any task
mm, you're right
Hrm, I see.
Hrm, actually, judging by their audit trail. Looks like they migrated 6 months ago. What do you guys think, try a re-clone, then if that doesn't work, just give instructions to push to new location?
<forgehg.hg_main.ForgeHgApp object="" at="" 0x2aaaacf2bb50=""> does not appear to be a cloned repo
I guess that means just pushing to new location.
I think I got that test figured out. Running it again just to make sure. Assuming everything goes well, just need to wait for the push to my fork to finish analyzing then I'll get a merge request in.