3842: rbtools should use the requests module instead of depending on urllib to do things
- WontFix
- RBTools
| yaneu******@gmai***** (Google Code) (Is this you? Claim this profile.) | |
What version are you running? 0.7.2 While trying to deconstruct what's going wrong for https://code.google.com/p/reviewboard/issues/detail?id=3840 , I'm more convinced that the problem with the original issue is that rbtools doesn't completely setup SSL certificates/information properly, which causes issues down the line if rbtools can't verify an ssl cert. Furthermore, there's a lot of complexity added to deal with opening URLs, generalizing connections, etc that don't need to be done in rbtools as it adds unnecessary complexity to the package. The requests module does a pretty awesome job at working with HTTP requests and the API doesn't change all that often. I highly recommend switching the underlying code from six to requests.
We've evaluated requests before. It might be worth re-evaluating it before RBTools 1.0 at some point. The reason we didn't go with requests in the past is that they had much narrower Python compatibility support than we do. This is not currently the case (we both have a minimum of Python 2.6), but that can of course change at any point. Assuming they're planning to be on Python 2.7 long-term, and assuming we decide to move to 2.7 as a minimum, we can move to using requests. That is, however, a large project, as we have a lot of infrastructure we'd need to carefully port over and thoroughly test.
-
- reviewboard + rbtools
-
- New + WontFix