Martin von Zweigbergk <martinvonz@google.com> [Fri, 10 Jan 2020 15:47:39 -0800] rev 44053
rebase: delete seemingly unnecessary needupdate()
This seemed to be about checking that the user hasn't updated away
when we asked them to resolve merge conflicts. These days we call
`cmdutil.checkunfinished()` and refuse to update, so the user
shouldn't be able to get into this state.
`test-rebase-interruptions.t` actually has some tests where it
disables the rebase extension in order to be allowed to do some of
these updates. That still passes, but I wouldn't personally haved
cared if that failed.
Differential Revision: https://phab.mercurial-scm.org/D7825
Martin von Zweigbergk <martinvonz@google.com> [Fri, 10 Jan 2020 13:24:25 -0800] rev 44052
workingctx: move setparents() logic from localrepo to mirror overlayworkingctx
It would be nice to later be able to call `wctx.setparents()` whether
`wctx` is a `workingctx` or an `overlayworkingctx`.
Differential Revision: https://phab.mercurial-scm.org/D7823
Martin von Zweigbergk <martinvonz@google.com> [Fri, 10 Jan 2020 21:41:28 -0800] rev 44051
overlayworkginctx: implement a setparents() to mirror dirstate.setparents()
This lets us make the in-memory and on-disk code a bit more
similar. I'll soon also implement setparents() on the regular
workingctx.
Differential Revision: https://phab.mercurial-scm.org/D7822
Martin von Zweigbergk <martinvonz@google.com> [Fri, 10 Jan 2020 17:03:23 -0800] rev 44050
overlayworkingctx: default branch to base context's branch
This matches what the dirstate does (it reuses working copy parent's
branch unless told otherwise). By moving the default out of
`rebase.commitmemorynode()`, it will let us clean that up better
later.
Differential Revision: https://phab.mercurial-scm.org/D7821