view tests/test-casefolding.t @ 47010:76ae43d5b1db stable

repoview: fix memory leak of filtered repo classes

The leak occurs in long-running server processes with
extensions, and is measured at 110kB per request.

Before this change, the contents of the `_filteredrepotypes`
cache are not properly garbage collected, despite it begin
a `WeakKeyDictionary`.

Extensions have a tendency to generate a new repository class
for each `localrepo` instantiation. Server processes based
on `hgwebdir_mod` will instantiate a new `localrepo` for each
HTTP request that involves a repository.

As a result, with a testing process that repeatedly opens a
repository with several extensions activated
(`topic` notably among them), we see a steady increase in
resident memory of 110kB per repository instantiation before this
change. This is also true, if we call `gc.collect()` at each
instantiation, like `hgwebdir_mod` does, or not.

The cause of the leak is that the *values* aren't weak references.
This change uses `weakref.ref` for the values, and this makes
in our measurements the resident size increase drop to 5kB per
repository instantiation, with no explicit call of `gc.collect()`
at all.
There is currently no reason to believe that this remaining leak
of 5kB is related to or even due to Mercurial core.

We've also seen evidence that `ui.ui` instances weren't properly
garbage collected before the change (with the change, they are).
This could explain why the figures are relatively high.

In theory, the collection of weak references could lead to
much more misses in the cache, so we measured the impact on
the original case that was motivation for introducing that cache
in 7e89bd0cfb86 (see also issue5043): `hg convert` of the
mozilla-central repository. The bad news here is that there is a
major memory leak there, both with and without the present changeset.
There were no more cache misses, and we could see no
more memory leak with this change: the resident size after importing
roughly 100000 changesets was at 12.4GB before, and 12.5GB after.
The small increase is mentioned for completeness only, and we
believe that it should be ignored, at least as long as the main
leak isn't fixed. At less than 1% of the main leak, even finding out
whether it is merely noise would be wasteful.

Original context where this was spotted and first mitigated:
  https://foss.heptapod.net/heptapod/heptapod/-/issues/466

The leak reduction was also obtained in Heptapod inner HTTP server,
which amounts to the same as `hgwebdir_mod` for these questions.

The measurements done with Python 3.9, similar figures seen with 3.8.
More work on our side would be needed to give measurements with 2.7,
because of testing server process does not support it.

#require icasefs

  $ hg debugfs | grep 'case-sensitive:'
  case-sensitive: no

test file addition with bad case

  $ hg init repo1
  $ cd repo1
  $ echo a > a
  $ hg add A
  $ hg st
  A a
  $ hg ci -m adda
  $ hg manifest
  a
  $ cd ..

test case collision on rename (issue750)

  $ hg init repo2
  $ cd repo2
  $ echo a > a
  $ hg --debug ci -Am adda
  adding a
  committing files:
  a
  committing manifest
  committing changelog
  updating the branch cache
  committed changeset 0:07f4944404050f47db2e5c5071e0e84e7a27bba9

Case-changing renames should work:

  $ hg mv a A
  $ hg mv A a
  $ hg st

addremove after case-changing rename has no effect (issue4590)

  $ hg mv a A
  $ hg addremove
  recording removal of a as rename to A (100% similar)
  $ hg revert --all
  forgetting A
  undeleting a

test changing case of path components

  $ mkdir D
  $ echo b > D/b
  $ hg ci -Am addb D/b
  $ hg mv D/b d/b
  D/b: not overwriting - file already committed
  ('hg rename --force' to replace the file by recording a rename)
  [1]
  $ hg mv D/b d/c
  $ hg st
  A D/c
  R D/b
  $ mv D temp
  $ mv temp d
  $ hg st
  A D/c
  R D/b
  $ hg revert -aq
  $ rm d/c
  $ echo c > D/c
  $ hg add D/c
  $ hg st
  A D/c
  $ hg ci -m addc D/c
  $ hg mv d/b d/e
  $ hg st
  A D/e
  R D/b
  $ hg revert -aq
  $ rm d/e
  $ hg mv d/b D/B
  $ hg st
  A D/B
  R D/b
  $ cd ..

test case collision between revisions (issue912)

  $ hg init repo3
  $ cd repo3
  $ echo a > a
  $ hg ci -Am adda
  adding a
  $ hg rm a
  $ hg ci -Am removea
  $ echo A > A

on linux hfs keeps the old case stored, force it

  $ mv a aa
  $ mv aa A
  $ hg ci -Am addA
  adding A

used to fail under case insensitive fs

  $ hg up -C 0
  1 files updated, 0 files merged, 1 files removed, 0 files unresolved
  $ hg up -C
  1 files updated, 0 files merged, 1 files removed, 0 files unresolved

no clobbering of untracked files with wrong casing

  $ hg up -r null
  0 files updated, 0 files merged, 1 files removed, 0 files unresolved
  $ echo gold > a
  $ hg up
  A: untracked file differs
  abort: untracked files in working directory differ from files in requested revision
  [20]
  $ cat a
  gold
  $ rm a

test that normal file in different case on target context is not
unlinked by largefiles extension.

  $ cat >> .hg/hgrc <<EOF
  > [extensions]
  > largefiles=
  > EOF
  $ hg update -q -C 1
  $ hg status -A
  $ echo 'A as largefiles' > A
  $ hg add --large A
  $ hg commit -m '#3'
  created new head
  $ hg manifest -r 3
  .hglf/A
  $ hg manifest -r 0
  a
  $ hg update -q -C 0
  $ hg status -A
  C a
  $ hg update -q -C 3
  $ hg update -q 0

  $ hg up -C -r 2
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
  $ hg mv A a
  $ hg diff -g > rename.diff
  $ hg ci -m 'A -> a'
  $ hg up -q '.^'
  $ hg import rename.diff -m "import rename A -> a"
  applying rename.diff
  $ hg st
  ? rename.diff
  $ hg files
  a
  $ find * | sort
  a
  rename.diff

  $ rm rename.diff

  $ cd ..

issue 3342: file in nested directory causes unexpected abort

  $ hg init issue3342
  $ cd issue3342

  $ mkdir -p a/B/c/D
  $ echo e > a/B/c/D/e
  $ hg add a/B/c/D/e
  $ hg ci -m 'add e'

issue 4481: revert across case only renames
  $ hg mv a/B/c/D/e a/B/c/d/E
  $ hg ci -m "uppercase E"
  $ echo 'foo' > a/B/c/D/E
  $ hg ci -m 'e content change'
  $ hg revert --all -r 0
  removing a/B/c/D/E
  adding a/B/c/D/e
  $ find * | sort
  a
  a/B
  a/B/c
  a/B/c/D
  a/B/c/D/e
  a/B/c/D/e.orig

  $ cd ..

issue 3340: mq does not handle case changes correctly

in addition to reported case, 'hg qrefresh' is also tested against
case changes.

  $ echo "[extensions]" >> $HGRCPATH
  $ echo "mq=" >> $HGRCPATH

  $ hg init issue3340
  $ cd issue3340

  $ echo a > mIxEdCaSe
  $ hg add mIxEdCaSe
  $ hg commit -m '#0'
  $ hg rename mIxEdCaSe tmp
  $ hg rename tmp MiXeDcAsE
  $ hg status -A
  A MiXeDcAsE
    mIxEdCaSe
  R mIxEdCaSe
  $ hg qnew changecase
  $ hg status -A
  C MiXeDcAsE

  $ hg qpop -a
  popping changecase
  patch queue now empty
  $ hg qnew refresh-casechange
  $ hg status -A
  C mIxEdCaSe
  $ hg rename mIxEdCaSe tmp
  $ hg rename tmp MiXeDcAsE
  $ hg status -A
  A MiXeDcAsE
    mIxEdCaSe
  R mIxEdCaSe
  $ hg qrefresh
  $ hg status -A
  C MiXeDcAsE

  $ hg qpop -a
  popping refresh-casechange
  patch queue now empty
  $ hg qnew refresh-pattern
  $ hg status
  $ echo A > A
  $ hg add
  adding A
  $ hg qrefresh a # issue 3271, qrefresh with file handled case wrong
  $ hg status # empty status means the qrefresh worked

#if osx

We assume anyone running the tests on a case-insensitive volume on OS
X will be using HFS+. If that's not true, this test will fail.

  $ rm A
  >>> open(u'a\u200c'.encode('utf-8'), 'w').write('unicode is fun') and None
  $ hg status
  M A

#endif

  $ cd ..