Re: Help needed to test packages with Django 1.7
On 08/03/2014 03:27 PM, Raphael Hertzog wrote:
> On Sun, 03 Aug 2014, Brian May wrote:
>>> Django 1.7 final isn't even released upstream, and therefore, downstream
>>> projects didn't even try to run against it. There *will* be issues we
>>> will have to deal with. 85 packages is quite something. I'm ok, and even
>>> welcome to *try* to make it before Jessie, though it is my view that
>>> it's unreasonable to rush without taking care of possible breaking.
>
> I already explained you in another thread on debian-release, that it's not
> reasonable security-wise to release Jessie with Django 1.6. And you're
> well placed to know what it means to maintain a package that gets security
> updates very often...
>
> We have very few Django end-users apps in Jessie, so let's try to take
> care of those but it's not a problem if some of the Django "extensions"
> (i.e. apps available for integration in custom developments that have
> almost no reverse dependencies) are dropped from jessie.
Well, currently, django 1.7 breaks Horizon. I've been trying to fix it,
and already found out that TEMPLATE_DIRS needed an additional comma in
settings.py, though there's still a lot more failure in the unit tests.
If you want to help, just try to rebuild Horizon (and add a comma as I
wrote in horizon/tests/settings.py to begin with...).
Horizon is my main concern.
>>> FYI, I'm trying to deal with python-memcache support for Python 3.4.
>>> Until that one is fixed, I wont be able to add support for Python 3 in
>>> keystoneclient, and therefore *a lot* of other packages will have no
>>> Python 3 support. I've tried to forward port a patch from
>>> python3-memcached, though it's still not ready, and unit tests are
>>> failing. Julien Danjou (eg: acid@d.o) wrote to me he'll try to find time
>>> to help. I really hope this one issue will be fixed soon, so that I can
>>> work on adding Python 3 everywhere possible in OpenStack, though right
>>> now it's a major blocker. I also hope we can upstream Python3 support
>>> for memcached, as the python3-memcached fork has been a major waste IMO.
>
> If you aren't able to deal with Django 1.7 support and Python 3 support
> before the freeze, please treat python 3 support as lower priority to Django
> 1.7 support. That said it would be nice to have both.
Well, I'm doing my best to add Python 3 support everywhere I can. I've
been doing this for months already. I know it wont be possible to fix
everything. Currently, I have 2 blockers which I am working on:
- python-memcache
- beautifulsoup
Both needs patches in upstream code, which I already partially wrote,
but it's not fully working (yet?). memcached isn't easy at all, as it
has to deal with lots of unicode stuff. Once these are done, then this
unlocks Python 3 support in maybe 40 other packages.
Anyway, this was to say: I don't think it's about one *or* the other,
I'm just trying to do my best to fix both, and it's not a question of
time (I do have time for it). Maybe I may try to get some help during
debconf from very knowledgeable people (Piotr and Barry comes to mind).
Cheers,
Thomas Goirand
Reply to: