At least for code review, the problem isn't pull requests as-such. The features that drive me to use other tools (on top of guthub hosting, often) are things like:

* The ability to mark which commits/files have been reviewed and so track progress.

* The ability to distinguish resolved and unresolved issues

* The ability to assign a reviewer based on which code is being altered in the PR.

* Useful handling of history rewriting during review.

I also think that the PR model could be implemented in a way that would make for better collaboration. At the moment, if someone submits a PR that has some issues and you want to help them fix those issues it's a huge pain because the commits typically live in their fork to which others usually don't have push access. It would be a considerable improvement if PR branches were set up in such a way that both the original submitter and the people with commit access to upstream could push to the review branch.

Beanstalk - private repo hosting for business - just released a brand new code review tool that we've been beta testing for a while now. It's rather nice and addresses many of the rough edges of pull requests. It's definitely for same-company teams as opposed to large open-source collaborative projects, though.

http://beanstalkapp.com

We host all of our private repos with Beanstalk, and use GH for our open source repos.

So, I'm no Jira fan either, but the code review side of github is definitely it's weakest point. They only just rolled out side by side reviews, something that I would have said was essential from day 1.

The fact that every comment on a review turns into a separate email is insanity, and just fills up your inbox with crap (one email for all of one person's comments on a review makes a ton more sense).

I hope they make it better, it definitely seems like a place where some TLC can go a long way.

Again, that's the idea with competition in the market. What one group wants may not be what everyone else wants. If enough people really want a specific set of features it would be nice if a vendor/Open Source solutions implemented that feature. There might for exmple be big differences in how enterprises use a tool and how (big) open source project use the same kind of tool.

I agree with that, but just to be clear, Stash is independent of Jira (though they do integrate) and has a simple pull request model. It's actually very nice to use (pretty much just like BitBucket).