view tests/test-config.t @ 51681:522b4d729e89

mmap: populate the mapping by default

Without pre-population, accessing all data through a mmap can result in many
pagefault, reducing performance significantly. If the mmap is prepopulated, the
performance can no longer get slower than a full read.
(See benchmark number below)

In some cases were very few data is read, prepopulating can be overkill and
slower than populating on access (through page fault). So that behavior can be
controlled when the caller can pre-determine the best behavior.
(See benchmark number below)

In addition, testing with populating in a secondary thread yield great result
combining the best of each approach. This might be implemented in later
changesets.

In all cases, using mmap has a great effect on memory usage when many processes
run in parallel on the same machine.

### Benchmarks

# What did I run

A couple of month back I ran a large benchmark campaign to assess the impact of
various approach for using mmap with the revlog (and other files), it
highlighted a few benchmarks that capture the impact of the changes well. So to
validate this change I checked the following:

- log command displaying various revisions
  (read the changelog index)
- log command displaying the patch of listed revisions
  (read the changelog index, the manifest index and a few files indexes)
- unbundling a few revisions
  (read and write changelog, manifest and few files indexes, and walk the graph
  to update some cache)
- pushing a few revisions
  (read and write changelog, manifest and few files indexes, walk the graph to
  update some cache, performs various accesses locally and remotely during
  discovery)

Benchmarks were run using the default module policy (c+py) and the rust one. No
significant difference were found between the two implementation, so we will
present result using the default policy (unless otherwise specified).

I ran them on a few repositories :

- mercurial: a "public changeset only" copy of mercurial from 2018-08-01 using
  zstd compression and sparse-revlog

- pypy: a copy of pypy from 2018-08-01 using zstd compression and sparse-revlog

- netbeans: a copy of netbeans from 2018-08-01 using zstd compression and
  sparse-revlog

- mozilla-try: a copy of mozilla-try from 2019-02-18 using zstd compression and
  sparse-revlog

- mozilla-try persistent-nodemap: Same as the above but with a persistent
  nodemap. Used for the log --patch benchmark only

# Results

For the smaller repositories (mercurial, pypy), the impact of mmap is almost
imperceptible, other cost dominating the operation. The impact of prepopulating
is undiscernible in the benchmark we ran.

For larger repositories the benchmark support explanation given above:

On netbeans, the log can be about 1% faster without repopulation (for a
difference < 100ms) but unbundle becomes a bit slower, even when small.

### data-env-vars.name                             = netbeans-2018-08-01-zstd-sparse-revlog
  # benchmark.name                                 = hg.command.unbundle
  # benchmark.variants.issue6528                   = disabled
  # benchmark.variants.reuse-external-delta-parent = yes
  # benchmark.variants.revs                        = any-1-extra-rev
  # benchmark.variants.source                      = unbundle
  # benchmark.variants.verbosity                   = quiet
with-populate: 0.240157
no-populate:   0.265087  (+10.38%, +0.02)
  # benchmark.variants.revs                        = any-100-extra-rev
with-populate: 1.459518
no-populate:   1.481290   (+1.49%, +0.02)
 ## benchmark.name                                 = hg.command.push
  # benchmark.variants.explicit-rev                = none
  # benchmark.variants.issue6528                   = disabled
  # benchmark.variants.protocol                    = ssh
  # benchmark.variants.reuse-external-delta-parent = yes
  # benchmark.variants.revs                        = any-1-extra-rev
with-populate: 0.771919
no-populate:   0.792025   (+2.60%, +0.02)
  # benchmark.variants.revs                        = any-100-extra-rev
with-populate: 1.459518
no-populate:   1.481290   (+1.49%, +0.02)


For mozilla-try, the "slow down" from pre-populate for small `hg log` is more
visible, but still small in absolute time. (using rust value for the persistent
nodemap value to be relevant).

### data-env-vars.name            = mozilla-try-2019-02-18-ds2-pnm
  # benchmark.name                = hg.command.log
  # bin-env-vars.hg.flavor        = rust
  # benchmark.variants.patch      = yes
  # benchmark.variants.limit-rev  = 1
with-populate: 0.237813
no-populate:   0.229452   (-3.52%, -0.01)
  # benchmark.variants.limit-rev  = 10
  # benchmark.variants.patch      = yes
with-populate: 1.213578
no-populate:   1.205189
### data-env-vars.name            = mozilla-try-2019-02-18-zstd-sparse-revlog
  # benchmark.variants.limit-rev  = 1000
  # benchmark.variants.patch      = no
  # benchmark.variants.rev        = tip
with-populate: 0.198607
no-populate:   0.195038   (-1.80%, -0.00)

However pre-populating provide a significant boost on more complex operations
like unbundle or push:

### data-env-vars.name                             = mozilla-try-2019-02-18-zstd-sparse-revlog
  # benchmark.name                                 = hg.command.push
  # benchmark.variants.explicit-rev                = none
  # benchmark.variants.issue6528                   = disabled
  # benchmark.variants.protocol                    = ssh
  # benchmark.variants.reuse-external-delta-parent = yes
  # benchmark.variants.revs                        = any-1-extra-rev
with-populate: 4.798632
no-populate:   4.953295   (+3.22%, +0.15)
  # benchmark.variants.revs                        = any-100-extra-rev
with-populate: 4.903618
no-populate:   5.014963   (+2.27%, +0.11)
 ## benchmark.name                                 = hg.command.unbundle
  # benchmark.variants.revs                        = any-1-extra-rev
with-populate: 1.423411
no-populate:   1.585365  (+11.38%, +0.16)
  # benchmark.variants.revs                        = any-100-extra-rev
with-populate: 1.537909
no-populate:   1.688489   (+9.79%, +0.15)

Windows needs ';' as a file separator in an environment variable, and MSYS
doesn't automatically convert it in every case.

#if windows
  $ path_list_var() {
  >     echo $1 | sed 's/:/;/'
  > }
#else
  $ path_list_var() {
  >     echo $1
  > }
#endif


hide outer repo
  $ hg init

Invalid syntax: no value

  $ cat > .hg/hgrc << EOF
  > novaluekey
  > EOF
  $ hg showconfig
  config error at $TESTTMP/.hg/hgrc:1: novaluekey
  [30]

Invalid syntax: no key

  $ cat > .hg/hgrc << EOF
  > =nokeyvalue
  > EOF
  $ hg showconfig
  config error at $TESTTMP/.hg/hgrc:1: =nokeyvalue
  [30]

Test hint about invalid syntax from leading white space

  $ cat > .hg/hgrc << EOF
  >  key=value
  > EOF
  $ hg showconfig
  config error at $TESTTMP/.hg/hgrc:1: unexpected leading whitespace:  key=value
  [30]

  $ cat > .hg/hgrc << EOF
  >  [section]
  > key=value
  > EOF
  $ hg showconfig
  config error at $TESTTMP/.hg/hgrc:1: unexpected leading whitespace:  [section]
  [30]

Reset hgrc

  $ echo > .hg/hgrc

Test case sensitive configuration

  $ cat <<EOF >> $HGRCPATH
  > [Section]
  > KeY = Case Sensitive
  > key = lower case
  > EOF

  $ hg showconfig Section
  Section.KeY=Case Sensitive
  Section.key=lower case

  $ hg showconfig Section -Tjson
  [
   {
    "defaultvalue": null,
    "name": "Section.KeY",
    "source": "*.hgrc:*", (glob)
    "value": "Case Sensitive"
   },
   {
    "defaultvalue": null,
    "name": "Section.key",
    "source": "*.hgrc:*", (glob)
    "value": "lower case"
   }
  ]
  $ hg showconfig Section.KeY -Tjson
  [
   {
    "defaultvalue": null,
    "name": "Section.KeY",
    "source": "*.hgrc:*", (glob)
    "value": "Case Sensitive"
   }
  ]
  $ hg showconfig -Tjson | tail -7
   {
    "defaultvalue": null,
    "name": "*", (glob)
    "source": "*", (glob)
    "value": "*" (glob)
   }
  ]

Test config default of various types:

 {"defaultvalue": ""} for -T'json(defaultvalue)' looks weird, but that's
 how the templater works. Unknown keywords are evaluated to "".

 dynamicdefault

  $ hg config --config alias.foo= alias -Tjson
  [
   {
    "name": "alias.foo",
    "source": "--config",
    "value": ""
   }
  ]
  $ hg config --config alias.foo= alias -T'json(defaultvalue)'
  [
   {"defaultvalue": ""}
  ]
  $ hg config --config alias.foo= alias -T'{defaultvalue}\n'
  

 null

  $ hg config --config auth.cookiefile= auth -Tjson
  [
   {
    "defaultvalue": null,
    "name": "auth.cookiefile",
    "source": "--config",
    "value": ""
   }
  ]
  $ hg config --config auth.cookiefile= auth -T'json(defaultvalue)'
  [
   {"defaultvalue": null}
  ]
  $ hg config --config auth.cookiefile= auth -T'{defaultvalue}\n'
  

 false

  $ hg config --config commands.commit.post-status= commands -Tjson
  [
   {
    "defaultvalue": false,
    "name": "commands.commit.post-status",
    "source": "--config",
    "value": ""
   }
  ]
  $ hg config --config commands.commit.post-status= commands -T'json(defaultvalue)'
  [
   {"defaultvalue": false}
  ]
  $ hg config --config commands.commit.post-status= commands -T'{defaultvalue}\n'
  False

 true

  $ hg config --config format.dotencode= format.dotencode -Tjson
  [
   {
    "defaultvalue": true,
    "name": "format.dotencode",
    "source": "--config",
    "value": ""
   }
  ]
  $ hg config --config format.dotencode= format.dotencode -T'json(defaultvalue)'
  [
   {"defaultvalue": true}
  ]
  $ hg config --config format.dotencode= format.dotencode -T'{defaultvalue}\n'
  True

 bytes

  $ hg config --config commands.resolve.mark-check= commands -Tjson
  [
   {
    "defaultvalue": "none",
    "name": "commands.resolve.mark-check",
    "source": "--config",
    "value": ""
   }
  ]
  $ hg config --config commands.resolve.mark-check= commands -T'json(defaultvalue)'
  [
   {"defaultvalue": "none"}
  ]
  $ hg config --config commands.resolve.mark-check= commands -T'{defaultvalue}\n'
  none

 empty list

  $ hg config --config commands.show.aliasprefix= commands -Tjson
  [
   {
    "defaultvalue": [],
    "name": "commands.show.aliasprefix",
    "source": "--config",
    "value": ""
   }
  ]
  $ hg config --config commands.show.aliasprefix= commands -T'json(defaultvalue)'
  [
   {"defaultvalue": []}
  ]
  $ hg config --config commands.show.aliasprefix= commands -T'{defaultvalue}\n'
  

 nonempty list

  $ hg config --config progress.format= progress -Tjson
  [
   {
    "defaultvalue": ["topic", "bar", "number", "estimate"],
    "name": "progress.format",
    "source": "--config",
    "value": ""
   }
  ]
  $ hg config --config progress.format= progress -T'json(defaultvalue)'
  [
   {"defaultvalue": ["topic", "bar", "number", "estimate"]}
  ]
  $ hg config --config progress.format= progress -T'{defaultvalue}\n'
  topic bar number estimate

 int

  $ hg config --config profiling.freq= profiling -Tjson
  [
   {
    "defaultvalue": 1000,
    "name": "profiling.freq",
    "source": "--config",
    "value": ""
   }
  ]
  $ hg config --config profiling.freq= profiling -T'json(defaultvalue)'
  [
   {"defaultvalue": 1000}
  ]
  $ hg config --config profiling.freq= profiling -T'{defaultvalue}\n'
  1000

 float

  $ hg config --config profiling.showmax= profiling -Tjson
  [
   {
    "defaultvalue": 0.999,
    "name": "profiling.showmax",
    "source": "--config",
    "value": ""
   }
  ]
  $ hg config --config profiling.showmax= profiling -T'json(defaultvalue)'
  [
   {"defaultvalue": 0.999}
  ]
  $ hg config --config profiling.showmax= profiling -T'{defaultvalue}\n'
  0.999

Test empty config source:

  $ cat <<EOF > emptysource.py
  > def reposetup(ui, repo):
  >     ui.setconfig(b'empty', b'source', b'value')
  > EOF
  $ cp .hg/hgrc .hg/hgrc.orig
  $ cat <<EOF >> .hg/hgrc
  > [extensions]
  > emptysource = `pwd`/emptysource.py
  > EOF

  $ hg config --source empty.source
  none: value
  $ hg config empty.source -Tjson
  [
   {
    "defaultvalue": null,
    "name": "empty.source",
    "source": "",
    "value": "value"
   }
  ]

  $ cp .hg/hgrc.orig .hg/hgrc

Test "%unset"

  $ cat >> $HGRCPATH <<EOF
  > [unsettest]
  > local-hgrcpath = should be unset (HGRCPATH)
  > %unset local-hgrcpath
  > 
  > global = should be unset (HGRCPATH)
  > 
  > both = should be unset (HGRCPATH)
  > 
  > set-after-unset = should be unset (HGRCPATH)
  > EOF

  $ cat >> .hg/hgrc <<EOF
  > [unsettest]
  > local-hgrc = should be unset (.hg/hgrc)
  > %unset local-hgrc
  > 
  > %unset global
  > 
  > both = should be unset (.hg/hgrc)
  > %unset both
  > 
  > set-after-unset = should be unset (.hg/hgrc)
  > %unset set-after-unset
  > set-after-unset = should be set (.hg/hgrc)
  > EOF

  $ hg showconfig unsettest
  unsettest.set-after-unset=should be set (.hg/hgrc)

Test exit code when no config matches

  $ hg config Section.idontexist
  [1]

sub-options in [paths] aren't expanded

  $ cat > .hg/hgrc << EOF
  > [paths]
  > foo = ~/foo
  > foo:suboption = ~/foo
  > EOF

  $ hg showconfig paths
  paths.foo=~/foo
  paths.foo:suboption=~/foo

note: The path expansion no longer happens at the config level, but the path is
still expanded:

  $ hg path | grep foo
  foo = $TESTTMP/foo

edit failure

  $ HGEDITOR=false hg config --edit
  abort: edit failed: false exited with status 1
  [10]

config affected by environment variables

  $ EDITOR=e1 VISUAL=e2 hg config --source | grep 'ui\.editor'
  $VISUAL: ui.editor=e2

  $ VISUAL=e2 hg config --source --config ui.editor=e3 | grep 'ui\.editor'
  --config: ui.editor=e3

  $ PAGER=p1 hg config --source | grep 'pager\.pager'
  $PAGER: pager.pager=p1

  $ PAGER=p1 hg config --source --config pager.pager=p2 | grep 'pager\.pager'
  --config: pager.pager=p2

verify that aliases are evaluated as well

  $ hg init aliastest
  $ cd aliastest
  $ cat > .hg/hgrc << EOF
  > [ui]
  > user = repo user
  > EOF
  $ touch index
  $ unset HGUSER
  $ hg ci -Am test
  adding index
  $ hg log --template '{author}\n'
  repo user
  $ cd ..

alias has lower priority

  $ hg init aliaspriority
  $ cd aliaspriority
  $ cat > .hg/hgrc << EOF
  > [ui]
  > user = alias user
  > username = repo user
  > EOF
  $ touch index
  $ unset HGUSER
  $ hg ci -Am test
  adding index
  $ hg log --template '{author}\n'
  repo user
  $ cd ..

configs should be read in lexicographical order

  $ mkdir configs
  $ for i in `$TESTDIR/seq.py 10 99`; do
  >    printf "[section]\nkey=$i" > configs/$i.rc
  > done
  $ HGRCPATH=configs hg config section.key
  99

Listing all config options
==========================

The feature is experimental and behavior may varies. This test exists to make sure the code is run. We grep it to avoid too much variability in its current experimental state.

  $ hg config --exp-all-known | grep commit | grep -v ssh
  commands.commit.interactive.git=False
  commands.commit.interactive.ignoreblanklines=False
  commands.commit.interactive.ignorews=False
  commands.commit.interactive.ignorewsamount=False
  commands.commit.interactive.ignorewseol=False
  commands.commit.interactive.nobinary=False
  commands.commit.interactive.nodates=False
  commands.commit.interactive.noprefix=False
  commands.commit.interactive.showfunc=False
  commands.commit.interactive.unified=None
  commands.commit.interactive.word-diff=False
  commands.commit.post-status=False
  convert.git.committeractions=[*'messagedifferent'] (glob)
  convert.svn.dangerous-set-commit-dates=False
  experimental.copytrace.sourcecommitlimit=100
  phases.new-commit=draft
  ui.allowemptycommit=False
  ui.commitsubrepos=False


Configuration priority
======================

setup necessary file

  $ cat > file-A.rc << EOF
  > [config-test]
  > basic = value-A
  > pre-include= value-A
  > %include ./included.rc
  > post-include= value-A
  > [command-templates]
  > log = "value-A\n"
  > EOF

  $ cat > file-B.rc << EOF
  > [config-test]
  > basic = value-B
  > [ui]
  > logtemplate = "value-B\n"
  > EOF


  $ cat > included.rc << EOF
  > [config-test]
  > pre-include= value-included
  > post-include= value-included
  > EOF

  $ cat > file-C.rc << EOF
  > %include ./included-alias-C.rc
  > [ui]
  > logtemplate = "value-C\n"
  > EOF

  $ cat > included-alias-C.rc << EOF
  > [command-templates]
  > log = "value-included\n"
  > EOF


  $ cat > file-D.rc << EOF
  > [command-templates]
  > log = "value-D\n"
  > %include ./included-alias-D.rc
  > EOF

  $ cat > included-alias-D.rc << EOF
  > [ui]
  > logtemplate = "value-included\n"
  > EOF

Simple order checking
---------------------

If file B is read after file A, value from B overwrite value from A.

  $ HGRCPATH=`path_list_var "file-A.rc:file-B.rc"` hg config config-test.basic
  value-B

Ordering from include
---------------------

value from an include overwrite value defined before the include, but not the one defined after the include

  $ HGRCPATH="file-A.rc" hg config config-test.pre-include
  value-included
  $ HGRCPATH="file-A.rc" hg config config-test.post-include
  value-A

command line override
---------------------

  $ HGRCPATH=`path_list_var "file-A.rc:file-B.rc"` hg config config-test.basic --config config-test.basic=value-CLI
  value-CLI

Alias ordering
--------------

The official config is now `command-templates.log`, the historical
`ui.logtemplate` is a valid alternative for it.

When both are defined, The config value read the last "win", this should keep
being true if the config have other alias. In other word, the config value read
earlier will be considered "lower level" and the config read later would be
considered "higher level". And higher level values wins.

  $ HGRCPATH="file-A.rc" hg log -r .
  value-A
  $ HGRCPATH="file-B.rc" hg log -r .
  value-B
  $ HGRCPATH=`path_list_var "file-A.rc:file-B.rc"` hg log -r .
  value-B

Alias and include
-----------------

The pre/post include priority should also apply when tie-breaking alternatives.
See the case above for details about the two config options used.

  $ HGRCPATH="file-C.rc" hg log -r .
  value-C
  $ HGRCPATH="file-D.rc" hg log -r .
  value-included

command line override
---------------------

  $ HGRCPATH=`path_list_var "file-A.rc:file-B.rc"` hg log -r . --config ui.logtemplate="value-CLI\n"
  value-CLI