Hey everybody. While this was supposed to be an offline and online experience I did not manage to get a space organized that would work out for the 18th which will make this an online only event.
There are two immediate goals for it:
- Redo the Jinja2 port to not use 2to3. This is of high interest because currently one of the reasons not much changes in Jinja2 is that I'm afraid I break the Python 3 port.
- Port most of Werkzeug to 3. There are already some patches in Ronny's repository and we should be able to build at least a basic version that uses that. This will also mean some minor API changes for Werkzeug on Python 2 due to how some of the interface do not survive Python 3 otherwise.
The coordination for this will happen in the #pocoo-sprint IRC
channel on irc.freenode.net. I'm in there with the mitsuhiko nickname
as you would expect. To join all you need is an IRC client and connect to
that channel.
If you want to join fill out this form real quick: bit.ly/join-the-sprint.
Not much really, everybody is welcome to join in. It's online online this time around so don't be shy. Ideling is allowed as well :)
If you don't think you're up for it still join, you can also help out just going through pull requests and documentation.
In any case git experience is a very good thing to have, ideally you also have a github account to make sharing things easy.
I will lay out a list of tickets and tag them appropriately for how we're going to tackle things. Generally it depends on the project, here is the plan:
Werkzeug:
Ronny's port generally is good enough as a starting point. We will however need to verify that all the API did not take a huge hit and performs as it should do.
We will start with looking at the port and verify testcase fo testcase if it makes sense.
The Werkzeug documentation is pretty bad in general, some cleanup there will be necessary and a new chapter on Python 3 is required.
Jinja2:
This one is tricky and will probably be a two step approach. Jinja2 currently uses 2to3 for the Python 3 support and it's pretty ugly. The idea would be to use python-modernize, a version of 2to3 that spits out code that works with 2.x and 3.x.
I did some tests on it recently and there are some problems with it. Maybe it would make sense to try to improve python-modernize before we tackle that thing.
Flask:
- Step one would be to go through the codebase with python-modernize and fix the issues that it leaves behind. Ideally we have a codebase that would import by itself on Python 2 and Python 3. The code is already written in a way that it does not require much porting.
- In theory when we do that we should once Werkzeug catches up have a version of Flask that runs on Python 3
I am aiming for June 1st to have an offline sprint in London if people are interested in it. There will be stuff remaining after Saturday and it only makes sense to have an offline one. If that happens I will sponsor some pizzas and beer for the cause :-)
@ThomasWaldmann I can say for Werkzeug it is Python 3.3+, i guess Flask won't be any different then.