DjangoBot: #django-dev

why is http://code.djangoproject.com/browser/django/tr... only checking for a single underscore when really it should be checking for double underscores? am i missing something here?

...Good question.

my ModelAdmin validation tests catch it. must have been an oversight due to the face that code isn't tested

s/face/fact/

Hooray for unit tests finding bugs.

:)

Newmaniese: you see my comment?

brosner: nope must've missed it, where at?

just posted it

hmm, didn't get to my email somehow. let me wade back into that. Thanks for looking at it

yeah want to get it done tonight

brosner: I moved the delete test cookie up because both of the two cases have test cookies that should be deleted. I am now sure why I used is_authenticated as opposed to is_active. Looking into that a little deeper

brosner: the patch doesn't apply cleanly now any way. Should I include those international exception formatting/ugettext?

ideally, don't include anything not directly apart of making what simon did better

it doesn't seem either of those things i mentioned are directly related are they?

nope definitely not, i'll take them out

thanks

you are welcome to file a new ticket if they are problems. just describe them well

:)

yea, I will leave that in Malcolm or Simon's hands, I only follow what I think are the rules, I really have never actually needed translations stuff. I am sure that stuff will get ironed out quickly.

brosner: I am looking at the is_authenticated and i got that from the has_permission function

even though it was is_active before

Should the has_permission check for is_active?

no it only needs to be checked on login

well maybe not

at this point that has always been the case so dont worry about it. just file a ticket

unless someone already did (likely)

Ok, just with default admin that shouldn;t be a problem but if there were custom login functions there could be a security issue. but yea it is in the old admin

i am thinking the patch that calls self.root is the correct way. is there any problems that might crop up re-running those checks prior to the self.login call?

any possibility self.login gets call again?

and it gets into a loop? I don't think so because it checks in the root function for is_authenticated (always true if there is a user after login) and is_staff

if the user isn't staff they get spit out before it recalls root

I just uploaded a new patch that applies cleanly, gets rid of the extra stuff and uses is_active instead

I agree with the root function because it is essentially the same request that was going through before the session expired or whatever

ok last question

err two

so deleting the test cookie earlier is fixing a problem related to this?

i don't see it as such

it wasn't a problem. it just left the test cookie there if there was post data

and last question, what happens when self.root_path goes away?

I was looking at that over the weekend, with the patch Alex Gaynor and I were working on. That's a good question.

does it make the exception one a little bit more forward thinking

We could just implement the same system that the admin site now uses with a decorator. I don't think the exception is much better because the error handling is still in the root fuction

alright well since that is likely post-1.0 i'll just commit this

Looking at the patch Alex and I did we could copy the current admin code almost verbatum. I agree though it would be too diffiuclt to try to fix this for forward compatiblity right now because in the root function we are not aware of the view that would be served if the user had been properly logged in

brosner: this might be too late, I am just looking at auth code and realizing when auth.login is called the last_login is set. We can get rid of the two lines that do that in the admin login section

I can open a ticket if it is too late

open a ticket

and didn't realize you said nearly 30 minutes ago ;)

brosner: All good! I wanted to thank you one more time for your work on NFA. It is pretty fricken awesome and I know you put a ton of time into it. I, on a very personal level, really appreciate it.

thanks Newmaniese. i just want to see this sucker merged :) you going to DjangoCon?

I can't wait to go back to using django-trunk:) I would love to get to DjangoCon. I need to convince my bosses for a plane ticket to Calif. (if I can secure a ticket first of course) I am getting married in a couple of weeks and have no extra money :(

ah hopefully it works out. and congratulations on the upcoming marriage :)

Thanks man

Newmaniese: congrats man. Want to borrow my "Just Married" shirt? I'm not using it these days. :P

empty: sure, thanks for the offer! It's getting a little intimidating right now and of course I have to get a site launched between now and then ...

hehe

Sorry if i ask this here but I've been asking this for days in the users channel with no luck!

I have a xmlfield which i want to filter somewhat before presenting it in the admin/change_list_results.html page. So basically i want to alter the string representation of this field. What is the place where I'm supposed to do this?

I just read the nfa status update, that's great!

david`bgk: yep I think we all owe the NewForms-Admin team a bunch of beer at djangocon. :)

:)

well, it'll probably be a virtual one from me but congratualitions to the team :)

congratulations*

maby I should check if I can send a case of beer with ups

jezdez: cheaper to call a local liquor store and have it delivered :P

hah, good point

but surely they dont have the good/rare german beer... ;)

did you hear that InBev bough Anheuser-Busch?

s/bough/bought/

oh wait, we are in #django-dev

sorry ^^

bdd

indeed, lol

does this look correct for newforms-admin dealing with inner Admin classes, http://dpaste.com/65373/ ?

jkocherhans: i am looking at #5609 and wondering if this even worth it? those classes were coming from how oldforms rendered fields.

brosner: I think it's worth fixing, but it sounds like a PITA

yeah...

;)

I think a lot of admin hackers are relying on those classes

but that said, it probably isn't *entirely* and nfa problem... a huge part of it is just newforms

s/and/an/

yeah. maybe it can be a known issue going into 1.0-alpha that we should definitely mark as pre-1.0

jkocherhans: i think you came in after i asked about http://dpaste.com/65373/ -- what do you think?

brosner: do you know if there's any precedent for linking to docs in error messages? seems like it would be *really* helpful in this case.

yeah i was thinking the same thing, but wasn't sure about that

i only see this being in there for the pre-1.0 releases

so i wouldn't think it would be a bad thing

what happens at 1.0? just no error?

i would think so.

or maybe its worth for a whole release?

When the excellent extensible command structure was implemented, was __call__ considered instead of handle?

HenrikV: I think it was considered but then decided for a handle_<command-type> style methods to make it more readable