Before, our API was to pass in a changectx. This was not as flexible as it could be. Let's add a new API to accept a node and rewrite pushfromchangeset() to call it. Accepting a hex node would have been easier because everything pushlog stores hex nodes. However, this is a bad design of pushlog: we should operate with binary nodes wherever possible. I felt like drawing a line in the sand. This change does break some filelog tests in hgext/hgmo for an obscure reason. The next commit will fix that. Review commit: https://reviewboard.mozilla.org/r/142094/diff/#index_header See other reviews: https://reviewboard.mozilla.org/r/142094/

This class was needed so we could pass a changectx like object into pushlog.pushfromchangeset(). Now that the pushlog has an API to accept a node, this fake API class isn't needed. Review commit: https://reviewboard.mozilla.org/r/142096/diff/#index_header See other reviews: https://reviewboard.mozilla.org/r/142096/

We have a custom wrapper function for the "filelog" web command that adds pushlog info. This was added in 3ac7f2de1e6b. The previous implementation was a bit fragile in that it relied on implementation details of the templater too much. We constructed a fake templater class and passed it into the original function. Then, we inspected the arguments passed to __call__, modified them in place, then called the original templater with those arguments. Something - I'm not sure what - broke in 4.2. This commit refactors the code to instead create a new proxy class for the templater that intercepts __call__ and does modification there. This fixes whatever the problem was with 4.2. While we could probably leave the .__class__ monkeypatched, it is restored out of paranoia. Review commit: https://reviewboard.mozilla.org/r/142098/diff/#index_header See other reviews: https://reviewboard.mozilla.org/r/142098/

We should be running the latest available Mercurial release in the hgdev environment to ensure things run with the latest version. Review commit: https://reviewboard.mozilla.org/r/142100/diff/#index_header See other reviews: https://reviewboard.mozilla.org/r/142100/

We've loosened our Mercurial upgrade policy to allow taking the initial X.Y release 2 weeks after release provided it looks stable. There haven't been any major changes to the stable branch of the Mercurial repo related to the 4.2 release. So I think it is stable enough to go out on the hgweb servers. Both hg-dev and hg-reviewboard were updated because we'll need to update templates and the 4.2 templates aren't compatible with 4.1. This is a bit unfortunate. But it should be relatively low risk. There's a slight chance we'll break the reviewboard server extension. But we have tests for that and tests are happy with this commit, so everything should be fine. The system package and virtualenvs for tools and replication are still on 4.1. We'll upgrade those to 4.2 once we upgrade the rest of the world to 4.2. Review commit: https://reviewboard.mozilla.org/r/142102/diff/#index_header See other reviews: https://reviewboard.mozilla.org/r/142102/

Templates from the Mercurial 4.2 release have been synchronized into the repo. We still need to synchronize changes to gitweb with our local fork. Review commit: https://reviewboard.mozilla.org/r/142104/diff/#index_header See other reviews: https://reviewboard.mozilla.org/r/142104/

The changes are pretty straightforward. Commit message <div> now have a new CSS class so they are fixed width. And there are a handful of changes to filelog to support line ranger filtering and display. Review commit: https://reviewboard.mozilla.org/r/142106/diff/#index_header See other reviews: https://reviewboard.mozilla.org/r/142106/