cmdstate: introduce a "saver" contextmanager and use it in place of save() Previously, the state was only saved in some paths out of these functions. This can be problematic, if the user ctrl-c's (or `kill -9`'s) the process, or we exit out of `relocate` for anything besides the "expected" reason, we won't record that we were in the middle of an evolve. One of our users has discovered that this leaves hg in a weird state; the user did something like this: ``` $ hg evolve <something goes wrong with the merge tool, hits ctrl-c> <deals with the merge conflicts> $ hg evolve --continue abort: no interrupted evolve to continue $ hg evolve abort: uncommitted changes # Note: commands.status.verbose=True is set. $ hg status M foo # The repository is in an unfinished *update* state. # No unresolved merge conflicts # To continue: hg update ``` The user did an `hg update`, but it didn't actually do anything besides take it out of the unfinished update state (the files were still dirty in the working directory).

cmdstate: avoid setting a default argument to a mutable object (`{}`) If there's ever more than one cmdstate for the lifetime of the process, this can cause surprising behavior where the later cmdstates pick up options from the earlier ones. I've not seen any evidence this is actually causing any issues, but it's subtle enough that it should probably be fixed to help save significant debugging time later.

cmdstate: switch to new-style classes This isn't a huge issue and isn't necessary on Python3, but we're still Python2 compatible and we should be consistent (and strongly prefer new-style classes); this is the only class that is not deriving from object that I could find.

branching: merge stable into default

docs: add example for the `prune` command Giving at least one example helps clear up the syntax, especially since most options will follow a very similar pattern.

tests: add some more actions to test-issue-6028, describe what's being done This patch does two things: it explains what's happening in the test file so it's easier to understand, and also it adds checks that make sure something like swapping merge parents is not accidentally breaking anything. The primary reason to touch this test file was that it was broken by a change in core's merge.graft() that erroneously swapped merge parents. Since only evolve uses merge.graft() for merge commits and there aren't any tests in core for it, let's test it here. Plus, this test case is pretty simple and these additional checks don't make it too complicated.

topic: fix some API to make it more robust The new code can deal with changes in upstream mercurial.

rewriteutil: move allowdivergence check to foldcheck() foldcheck() calls precheck() first and then performs more fold-specific checks.

test: change to output around transaction timing CORE-TEST-OUTPUT-UPDATE: d7304434390f

branching: merge with stable There have been multiple fix on stable.

stablerange: add some data from field testing This show that in number of stable range to track stay reasonable.

stablerange: add a debugstablerangecache command The command display get some data about stable range in a repository.

stablerange: add some documentation about the general concept

stablesort: add some field data about stable sort cache

stablesort: introduce a small debugstablesortcache command This give a small insight on the cache content.

stablesort: add some documentation for stablesort This should help people to understand what is going on here.

fold: check allowdivergence before folding obsolete changesets (issue5817)

tests: demonstrate that fold doesn't care about allowdivergence config option

stack: remove unnecessary copying of rdependencies rdependencies is not modified in any way in this method, so no need to copy it.

stack: make a deep copy of `dependencies` before modifying its items The algorithm later on in this method uses .remove() to remove individual elements from items in dependencies, which before this patch modified the cached property contents. So for further use that dictionary was in the form of {1: set([])}, i.e. all sets were empty. This deep copy block could be way simpler, but the problem is that sometimes we get lists of _succs() from evolvebits.builddependencies(). Note: this happens only in topic's stack version of builddependencies() and it looks like a suboptimal way to handle multiple successors (see evolve's counterpart function). stack.builddependencies method is removed, it has served its purpose (see the previous patch).

stack: demonstrate that not reusing cached property gives different results Instead of using self._dependencies, which is a cached property, let's get fresh deps and rdeps in stack.revs method using a newly introduced stack.builddependencies method (will be removed in the next patch). This makes it easier to explain why the next patch is correct.

branching: merge with stable

changelog: add missing entry for 9af212b8565a

evolve: test that target is not orig in _solveunstable() (issue6097) `newer` is the result of obsutil.successorssets() and can be [[orig.node()]], in which case later on in this function evolve will try to rebase orig onto orig, which is not correct. So let's just check this particular case. This fix doesn't cover cases when successorssets() result contains orig.node() not at [0][0]. Such cases need tests.