DjangoBot: #django-dev

hopefully, we won't hold it up :)

I agree

Nice Work Alex_Gaynor I'll chat with you tomorrow morning

hello

cramm, unfortunately it doesn't fix the bug for me, I still got the same error

Hi guys, anyone running OS X and willing to test a script?. I'm trying to learn why a Django test (trying loading a template from an egg) fails

s/learn/digg/

cramm: you still looking for a mac tester?

Newmaniese: oh yes. http://dpaste.com/64198/

Newmaniese: that's a standalone copy of tests/regressiontest/templates/loaders.py and django/template/loaders/eggs.py

ok I'll run it quick

Newmaniese: we have a test templates that is failing both under win32 and mac os x. I found how to solve it under win32 but I can't understand why it keeps failing under os x

the failure itself can be triggered by runnning Django's ./runtests --verbosity=2 templates

Newmaniese: oh, forgot to ask this first: Do you have setuptools?

cramm: nope I am realizing that is failing right now

I can install them quick

Oh that would be great

ping me if you need more testers

ok

in [7751] Adrian modified these tests to not be execised if setuptools isn't present

cramm: I am getting an error in the script: http://dpaste.com/64201/

I forgot to delete that line. But the fact that you reach that point would mean the Django test doesn't fail on your system

Weird

python 2.5 mac osx 10.5.4 and I just got the latest version of setup tools

Should I delete the return? just to see if the thing can finish

someone else also running reports it fails for him: http://dpaste.com/59930/

Newmaniese: yes please

cramm: another error: http://dpaste.com/64202/

because it isn;t returning anything :)

yeap :), I made a mess of that part

There we go: does this look right: http://dpaste.com/64216/

I just uncommented the line 30

Newmaniese: Thanks. So I now need to understand why is doesn't fails always. Will ask the reporter about the versions of pieces he is using, I only know he has setuptools 0.6c8

cramm: Yep that's the version I have. Ping me if there is any other way that I can help, I don't know much about eggs, but I do have a Mac :)

Ok, thanks again :)

in newforms-admin, will django.contrib.admin.autodiscover() be the prefered way to deal with auto-register?

Hey brosner

Arg, almost forgot to ask him!

Haha figured he was on might as well

Defniitely, it just slipped my mind

I am looking at the templates now and trying to figure out how to reverse the model_admin names

Hrm, yeah, we'd need to be able to do some sort of string iterpolation

I am thinking It might be better to pass the model stuff as args

You mean make a custom admin_url tag?

hey Newmaniese. i will be doing some django code in about an hour. mind pinging me then?

Sure Alex_Gaynor and I have some things for you to take a look at

Ok, I'm leaving in about 15 minutes, but Newmaniese knows what it is

Hey brosner. Now good?

yeah

Alex_Gaynor and I (mostly Alex_Gaynor) worked up a patch that fixes reverse in NFA. Only issue we are having is one test fails. It's the test where you monkey patch in the extra_context

This fails because we are using the traditional url_conf and so the urls and the methods are stored at the URL level. So monkey patching like that won't work

we messed around with it and we could create a lazy_loader or a factory function or we could just drop support for that style of monkey patching, which isn't supported anywhere else in django

hmm, are you sure you meant extra_context?

it's the way the admin views test does it on line 279

here is the still rough patch: http://dpaste.com/64358/

ideally to make this all work there needs to be some changes upstream in urlresolvers

We added one line to support objects passed into RegexURLResolver

but that's the only change, nothing else is affected

I guess the question is: If we want to continue support for items like that test that occur after runtime

Newmaniese: your patch has one very positive sideeffect, it's now really easy to add some custom views into the admin, just by overriding urls :)

thats always been possible apollo13 :)

until nfa changes

brosner: must have missed something :)

I don't follow you

you have always been able to override urls using urlpatterns since they short-curcuit, but nfa made that a bit harder

and technially that is still possible

brosner: of course, but that would remember them to put them before the admin include (if I understand you right), with this patch it would be tied to the adminsite/modeladmin, so there is no need for the user to edit anything

hmm let me write that one again^^

but that would require me to ...

apollo13: I think this method is a little more traditional, but adding views is one of the main goals of NFA

indeed

Newmaniese: not really. it would be a nice side-effect, but was never the intension since for now it will likely stay the same as it always has.

haha ok s/main goals/side effects/ :)

so brosner, what do you think?

i think its something ill spend more time looking at after a merge. want to get these other items done first

i am a little hesitant right now

about its approach

what are the side-effects of emulating a module?

you mean the url module?

types.ModuleType

Yea, ok, Alex_Gaynor would know more about that, but I have been testing in the shell and I am pretty sure it behaves the way you expect it to

cross vm compatible?

Haven't gone that far yet, but I will make sure to check before anything deemed usable is availible

cool

empty: you there?

siudesign: yep

hey siudesign

Tis Robert Lofthouse, DjangoCon chair. In response to your blog post, i've sent you a little email. Hope you're interested ;)

hey Brian

siudesign: thanks. I'll check it now. :)

good to hear it got officially announced :)

Yeah, finally! It's been a lot of fun, considering I live in London... and everyone else is in the US :) It's all worth it though

brosner: Obviously we need you to look at this deeply and I am sure you are busy right now. If we continued this route would it be bad to have that one test changed?