| FazBrowse GitHub Viewer | Trending | | Home |
| Tools: [Download Repo ZIP] [Original HTTPS Page] |
|
Changes Unknown when pulling f41eb74 on foozlevazquez:fix-header-strings-for-httplib2 into * on google:master*. |
Sorry, something went wrong.
|
isn't this the same issue as #104 ? |
Sorry, something went wrong.
|
@craigcitro #104 is the same root problem, but the conclusion that the bug lies in httplib2 is mistaken: httplib2 clearly expects strings and not bytes in headers, so this is not a bug with httplib2. |
Sorry, something went wrong.
|
I would argue #104 is "on the right track" and that httplib2 should handle this. Uses of str to mean bytes in Py2 and to mean unicode in Py3 is just asking for bugs. |
Sorry, something went wrong.
|
Should I expect this to be solved soon or should I find a workaround? |
Sorry, something went wrong.
|
@cganterh This is only needed for Python 3, if that helps. |
Sorry, something went wrong.
|
@dhermes my whole project is based on Python 3 and is strongly dependent on Python 3 libraries. Is it a good idea to fork oauth2client or httplib2 and apply the pull requests myself so I can go on while this issue is being fixed? |
Sorry, something went wrong.
|
@cganterh: Why fork httplib2? What change would you want to make to httplib2 that hasn't been accepted by the httplib2 maintainers? |
Sorry, something went wrong.
|
Correct me if im wrong, but I think that #136, #104 and jcgregorio/httplib2#291 are trying to solve the same issue. However I haven't read the pull requests in depth. |
Sorry, something went wrong.
|
Okay, I suspected that they were all the same thing. It looks like getting a fix into httplib2 is blocked on writing test coverage for the fix, which is something that anyone can do. |
Sorry, something went wrong.
|
Yes, I know. But I think I'm not the most appropriate person to do this, since I have little experience on writing tests and it is my first time using the library. So I thought it was a bad idea to do it myself and maybe a better one was to just use some of the pull requests while the problem was actually being fixed. Now I think it doesn't make sense to fork the whole library rather than using the fork that the author of the pull request already made. |
Sorry, something went wrong.
|
@anthmgoogle I don't think @jcgregorio is actively maintaining httplib2, so it may be best to just get the support in oauth2client. Maybe I am mistaken though. |
Sorry, something went wrong.
|
All that's stopping httplib2 from having the change is that the change has to come with unit test coverage. That's not-at-all an unmaintained project. :-) |
Sorry, something went wrong.
|
So I finally decided to write the unit test for jcgregorio/httplib2#291. |
Sorry, something went wrong.
Sorry, something went wrong.
|
What action remains to be taken in this pull request now that the fix in httplib2 has been made? |
Sorry, something went wrong.
|
@nathanielmanistaatgoogle I am going through this to see if there is anything we missed in #250 |
Sorry, something went wrong.
| Back | FazBrowse Home | New Git URL |
It appears that earlier contributors made the attempt to keep everything as bytes, mostly deducing from the test code. This doesn't work with the Python3 version of httplib2, and it is exactly this issue (bytes being passed to the Python3 version of httplib2 instead of str) that causes corrupt headers. httplib2 is very clear about what it expects.
Given the dependency on httplib2, in oauth2client, it makes sense to me that oauth2client should follow its lead and use whatever version of str for headers makes sense in the version of Python it is using.