push: rework the computation of fallbackheads to be correct
The previous computation tried to be smart but ended up being wrong. This was
caught by phase movement test while reworking the phase discovery logic to be
faster.
The previous logic was failing to catch case where the pushed set was not based
on a common heads (i.e. when the discovery seemed to have "over discovered"
content, outside the pushed set)
In the following graph, `e` is a common head and we `hg push -r f`. We need to
detect `c` as a fallback heads and we previous failed to do so::
e
|
d f
|/
c
|
b
|
a
The performance impact of the change seems minimal. On the most impacted
repository at hand (mozilla-try), the slowdown seems mostly mixed in the
overall noise `hg push` but seems to be in the hundred of milliseconds order of
magnitude. When using rust, we seems to be a bit faster, probably because we
leverage more accelaratd internals.
I added a couple of performance related common for further investigation later
on.
#testcases dirstate-v1 dirstate-v2
#if dirstate-v2
$ cat >> $HGRCPATH << EOF
> [format]
> use-dirstate-v2=1
> [storage]
> dirstate-v2.slow-path=allow
> EOF
#endif
https://bz.mercurial-scm.org/660 and:
https://bz.mercurial-scm.org/322
$ hg init repo
$ cd repo
$ echo a > a
$ mkdir b
$ echo b > b/b
$ hg commit -A -m "a is file, b is dir"
adding a
adding b/b
File replaced with directory:
$ rm a
$ mkdir a
$ echo a > a/a
Should fail - would corrupt dirstate:
$ hg add a/a
abort: file 'a' in dirstate clashes with 'a/a'
[255]
Removing shadow:
$ hg rm --after a
Should succeed - shadow removed:
$ hg add a/a
Directory replaced with file:
$ rm -r b
$ echo b > b
Should fail - would corrupt dirstate:
$ hg add b
abort: directory 'b' already in dirstate
[255]
Removing shadow:
$ hg rm --after b/b
Should succeed - shadow removed:
$ hg add b
Look what we got:
$ hg st
A a/a
A b
R a
R b/b
Revert reintroducing shadow - should fail:
$ rm -r a b
$ hg revert b/b
abort: file 'b' in dirstate clashes with 'b/b'
[255]
Revert all - should succeed:
$ hg revert --all
forgetting a/a
forgetting b
undeleting a
undeleting b/b
$ hg st
Issue3423:
$ hg forget a
$ echo zed > a
$ hg revert a
$ hg st
? a.orig
$ rm a.orig
addremove:
$ rm -r a b
$ mkdir a
$ echo a > a/a
$ echo b > b
$ hg addremove -s 0
removing a
adding a/a
adding b
removing b/b
$ hg st
A a/a
A b
R a
R b/b
commit:
$ hg ci -A -m "a is dir, b is file"
$ hg st --all
C a/a
C b
Long directory replaced with file:
$ mkdir d
$ mkdir d/d
$ echo d > d/d/d
$ hg commit -A -m "d is long directory"
adding d/d/d
$ rm -r d
$ echo d > d
Should fail - would corrupt dirstate:
$ hg add d
abort: directory 'd' already in dirstate
[255]
Removing shadow:
$ hg rm --after d/d/d
Should succeed - shadow removed:
$ hg add d
$ hg ci -md
Update should work at least with clean working directory:
$ rm -r a b d
$ hg up -r 0
2 files updated, 0 files merged, 0 files removed, 0 files unresolved
$ hg st --all
C a
C b/b
$ rm -r a b
$ hg up -r 1
2 files updated, 0 files merged, 0 files removed, 0 files unresolved
$ hg st --all
C a/a
C b