Re: GSOC 2015 project ideas suggestion

2015-02-21 Thread Russell Keith-Magee
To my mind, the goal of this project isn't to replace Django's routing infrastructure - as Josh and Florian have pointed out, unless you can provide a backwards compatible interface as well as tangible API improvements, we're unlikely to adopt that code. However, there *would* be benefit in making

Re: GSOC 2015 project ideas suggestion

2015-02-21 Thread Josh Smeaton
To expand on Florians reply, why do you think replacing the routing infrastructure is a good idea? Is there any tangible benefit in doing so? Can you demonstrate that you can stay completely backwards compatible while realising those benefits? If there answer is no to benefits or backwards comp

Re: status of 1.8 release blockers

2015-02-21 Thread Tim Graham
Patches for the two remaining blockers are awaiting review. #24381 Cache pickling exception in 1.8a1 with cross-table filter params #24377 UUIDField as primar

Re: GSOC 2015 project ideas suggestion

2015-02-21 Thread Florian Apolloner
Hi Aisf, while it theoretically would be possible to replace all of Django's request/response handling with Werkzeug, there is not much gain in it currently -- especially when considering backwards compatibility etc… Cheers, Florian On Saturday, February 21, 2015 at 9:14:27 PM UTC+1, Asif Saif

Re: GSOC 2015 project ideas suggestion

2015-02-21 Thread Asif Saifuddin
Hi, Analyzing django code base docs and other tools I'm thinking of using werkzeug's routing infrastructure for developing django's one. django http mechanism can e re written using webOb/werkzeugs utilities too. this will need working django urlresolve, urls, middlewares, views, http and some mor

Re: Should the test suite pass on a read-only filesystem?

2015-02-21 Thread Claude Paroz
Le samedi 21 février 2015 12:08:38 UTC+1, Aymeric Augustin a écrit : (...) > I foresee two difficulties: > > - How do we enforce this in the long run? Can we run the CI server with a > very > unprivileged user that isn’t allowed to write anywhere other than /tmp? > - makemessages isn’t flex

Re: Should the test suite pass on a read-only filesystem?

2015-02-21 Thread Tim Graham
Sounds good to me. On Jenkins, I think using chmod to remove write permissions from the relevant checkout directories might work. On Saturday, February 21, 2015 at 6:08:38 AM UTC-5, Aymeric Augustin wrote: > > Hello, > > Would it be a good property of the test suite to pass on a read-only > che

Should the test suite pass on a read-only filesystem?

2015-02-21 Thread Aymeric Augustin
Hello, Would it be a good property of the test suite to pass on a read-only checkout? Here’s one way to try it without messing up your git checkout (guess why I’m mentioning this): % find django tests -type f -print0 | xargs -0 chmod u-w % find django tests -type d -print0 | xargs -0 chmod u-w %