Py3k compatibility

[tseaver]

All tests pass under tox against Python 2.6, 2.7, 3.2, and 3.3.

[zagy]

Hi, how is that with merging and releasing for python3 compatibility? I'd need it. Anything I can do to help?

[zagy]

For the time being I've added a release of Tres' branch to http://download.gocept.com/packages/oauth2-1.5.211py3-1.zip

[coilysiren]

+1

[joestump]

See #158 as well. cc @jaitaiwan

[joestump]

@tseaver You interested in becoming a collaborator on this? Second solid PR I've seen from you so far. 👍

[jaitaiwan]

@joestump which PR should we use?

[jaitaiwan]

This one has collisions...

[joestump]

@jaitaiwan Up to you. I think yours is newer and had more eyes on it, which always gets my vote. I mainly wanted to ask if @tseaver was interested in helping out. 😆

[tseaver]

I had just about enough mental swap space left to fix #141. I'm afraid that updating this one requires swapping way more of the code back into my head than I have available for the foreseeable future.

[joestump]

@tseaver haha, totally understand. Thanks so much for your help with #141 and your PRs. 🎉

[hayd]

There's a subtley here IMO, should all the things return as utf8 encoded bytes (as is the idea in this PR) or unicode (as most of stdlib does).

In a trial merge I changed it to unicode and basically the only fail was test_to_url_nonascii. Thoughts?

[rickhanlonii]

@jaitaiwan @joestump I spent some time today trying to fix the tests on #176 and found that we're going to need to make rather extensive changes if we want to use that PR (i.e. in Python 3 urlparse.urlsplit returns bytes so string checks fail). Those changes seem to already be contained in this PR.

I'm thinking we get this merged into feat-python-3 so we can all work on getting python 3 compatibility working with this as a base.

[rickhanlonii]

I attempted a master on merge here.

All tests pass for 2.6 and 2.7. There are a only 2 of the new tests failing in python 3:

Failure
Traceback (most recent call last):
  File "/Users/euler/dev/python-oauth2/py3.4/lib/python3.4/site-packages/mock/mock.py", line 1305, in patched
    return func(*args, **keywargs)
  File "/Users/euler/dev/python-oauth2/tests/test_oauth.py", line 908, in test_request_nonutf8_bytes
    self.assertRaises(TypeError, oauth.Request, method="GET", url=url, parameters=params)
AssertionError: TypeError not raised by Request


Failure
Expected :'http[38 chars]asdf&foo=baz&uni_utf8=%C2%AE&bar=foo&multi=%5B[81 chars]3%98'
Actual   :'http[38 chars]asdf&uni_unicode_2=%C3%A5%C3%85%C3%B8%C3%98&un[79 chars]=baz'
 <Click to see difference>

Traceback (most recent call last):
  File "/Users/euler/dev/python-oauth2/tests/test_oauth.py", line 478, in test_to_url_works_with_non_ascii_parameters
    'http://example.com?oauth_consumer=asdfasdfasdf&'
AssertionError: 'http[38 chars]asdf&foo=baz&uni_utf8=%C2%AE&bar=foo&multi=%5B[81 chars]3%98' != 'http[38 chars]asdf&uni_unicode_2=%C3%A5%C3%85%C3%B8%C3%98&un[79 chars]=baz'
- http://example.com?oauth_consumer=asdfasdfasdf&foo=baz&uni_utf8=%C2%AE&bar=foo&multi=%5Bb%27FOO%27%2C+b%27BAR%27%5D&uni_unicode=%C2%AE&uni_unicode_2=%C3%A5%C3%85%C3%B8%C3%98
+ http://example.com?oauth_consumer=asdfasdfasdf&uni_unicode_2=%C3%A5%C3%85%C3%B8%C3%98&uni_utf8=%C2%AE&multi=%5B%27FOO%27%2C+%27BAR%27%5D&uni_unicode=%C2%AE&bar=foo&foo=baz

[jaitaiwan]

Getting closer!

[hayd]

for the latter test: %5Bb%27FOO%27%2C+b%27BAR%27%5D has the b in b'..' which I don't think it should. I think it's easiest just to split i on ? then & (and test the set).

Thoughts on the bytes vs unicode issue?

[jaitaiwan]

I personally opt for unicode, I feel like its the way of the future...

[hayd]

In which case (I have a merge) which fails only: test_to_url_nonascii (py2.6+ and py3) and test_to_url_works_with_non_ascii_parameters (py2). These are kinda tricky since the unicode arguments aren't valid unicode (so it feels like this is expected).

Edit: See test failures here: https://travis-ci.org/hayd/python-oauth2

[jaitaiwan]

Hey @hayd do you reckon you'd be interested in becoming a contributor? Seems like you're super keen to help this project along! cc @joestump

[rickhanlonii]

+1

[jaitaiwan]

cc @joestump: this is the PR we're looking at for py3k over my stuff.

[rickhanlonii]

@jaitaiwan @joestump should this be against develop or a feature branch?

[jaitaiwan]

It's a big change so I'd say feature branch

On Mon, 3 Aug 2015 11:27 Rick Hanlon II [email protected] wrote:

@jaitaiwan https://github.com/jaitaiwan @joestump
https://github.com/joestump should this be against develop or a feature
branch?

—
Reply to this email directly or view it on GitHub
#142 (comment)
.

[rickhanlonii]

@tseaver can you re-open against branch feat-python-3?

[tseaver]

@rickhanlonii I'm happy to have somebody rebase / merge my work to such a branch, but I'm afraid I don't have capacity for that effort at this point.

[rickhanlonii]

No sweat. @hayd and I both took a stab at it already, just wanted to give you the opportunity first.