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.
$ mkdir folder
$ cd folder
$ hg init
$ mkdir x x/l x/m x/n x/l/u x/l/u/a
$ touch a b x/aa.o x/bb.o
$ hg status
? a
? b
? x/aa.o
? x/bb.o
$ hg status --terse u
? a
? b
? x/
$ hg status --terse maudric
? a
? b
? x/
$ hg status --terse madric
? a
? b
? x/aa.o
? x/bb.o
$ hg status --terse f
abort: 'f' not recognized
[10]
Add a .hgignore so that we can also have ignored files
$ echo ".*\.o" > .hgignore
$ hg status
? .hgignore
? a
? b
$ hg status -i
I x/aa.o
I x/bb.o
Tersing ignored files
$ hg status -t i --ignored
I x/
Adding more files
$ mkdir y
$ touch x/aa x/bb y/l y/m y/l.o y/m.o
$ touch x/l/aa x/m/aa x/n/aa x/l/u/bb x/l/u/a/bb
$ hg status
? .hgignore
? a
? b
? x/aa
? x/bb
? x/l/aa
? x/l/u/a/bb
? x/l/u/bb
? x/m/aa
? x/n/aa
? y/l
? y/m
$ hg status --terse u
? .hgignore
? a
? b
? x/
? y/
Run from subdirectory
$ hg status --terse u --cwd x/l
? .hgignore
? a
? b
? x/
? y/
$ relstatus() {
> hg status --terse u --config commands.status.relative=1 "$@";
> }
This should probably have {"l/", "m/", "n/"} instead of {"."}. They should
probably come after "../y/".
$ relstatus --cwd x
? ../.hgignore
? ../a
? ../b
? .
? ../y/
This should probably have {"u/", "../m/", "../n/"} instead of {"../"}.
$ relstatus --cwd x/l
? ../../.hgignore
? ../../a
? ../../b
? ../
? ../../y/
This should probably have {"a/", "bb", "../aa", "../../m/", "../../n/"}
instead of {"../../"}.
$ relstatus --cwd x/l/u
? ../../../.hgignore
? ../../../a
? ../../../b
? ../../
? ../../../y/
This should probably have {"bb", "../bb", "../../aa", "../../../m/",
"../../../n/"} instead of {"../../../"}.
$ relstatus --cwd x/l/u/a
? ../../../../.hgignore
? ../../../../a
? ../../../../b
? ../../../
? ../../../../y/
$ hg add x/aa x/bb .hgignore
$ hg status --terse au
A .hgignore
A x/aa
A x/bb
? a
? b
? x/l/
? x/m/
? x/n/
? y/
Including ignored files
$ hg status --terse aui
A .hgignore
A x/aa
A x/bb
? a
? b
? x/l/
? x/m/
? x/n/
? y/l
? y/m
$ hg status --terse au -i
I x/aa.o
I x/bb.o
I y/l.o
I y/m.o
Committing some of the files
$ hg commit x/aa x/bb .hgignore -m "First commit"
$ hg status
? a
? b
? x/l/aa
? x/l/u/a/bb
? x/l/u/bb
? x/m/aa
? x/n/aa
? y/l
? y/m
$ hg status --terse mardu
? a
? b
? x/l/
? x/m/
? x/n/
? y/
Modifying already committed files
$ echo "Hello" >> x/aa
$ echo "World" >> x/bb
$ hg status --terse maurdc
M x/aa
M x/bb
? a
? b
? x/l/
? x/m/
? x/n/
? y/
Respecting other flags
$ hg status --terse marduic --all
M x/aa
M x/bb
? a
? b
? x/l/
? x/m/
? x/n/
? y/l
? y/m
I x/aa.o
I x/bb.o
I y/l.o
I y/m.o
C .hgignore
$ hg status --terse marduic -a
$ hg status --terse marduic -c
C .hgignore
$ hg status --terse marduic -m
M x/aa
M x/bb
Passing 'i' in terse value will consider the ignored files while tersing
$ hg status --terse marduic -u
? a
? b
? x/l/
? x/m/
? x/n/
? y/l
? y/m
Omitting 'i' in terse value does not consider ignored files while tersing
$ hg status --terse marduc -u
? a
? b
? x/l/
? x/m/
? x/n/
? y/
Trying with --rev
$ hg status --terse marduic --rev 0 --rev 1
abort: cannot use --terse with --rev
[10]
Config item to set the default terseness
$ cat <<EOF >> $HGRCPATH
> [commands]
> status.terse = u
> EOF
$ hg status -mu
M x/aa
M x/bb
? a
? b
? x/l/
? x/m/
? x/n/
? y/
Command line flag overrides the default
$ hg status --terse=
M x/aa
M x/bb
? a
? b
? x/l/aa
? x/l/u/a/bb
? x/l/u/bb
? x/m/aa
? x/n/aa
? y/l
? y/m
$ hg status --terse=mardu
M x/aa
M x/bb
? a
? b
? x/l/
? x/m/
? x/n/
? y/
Specifying --rev should still work, with the terseness disabled.
$ hg status --rev 0
M x/aa
M x/bb
? a
? b
? x/l/aa
? x/l/u/a/bb
? x/l/u/bb
? x/m/aa
? x/n/aa
? y/l
? y/m