view relnotes/5.1 @ 42743:8c9a6adec67a

rust-discovery: using the children cache in add_missing

The DAG range computation often needs to get back to very old
revisions, and turns out to be disproportionately long, given
that the end goal is to remove the descendents of the given
missing revisons from the undecided set.

The fast iteration capabilities available in the Rust case make
it possible to avoid the DAG range entirely, at the cost of
precomputing the children cache, and to simply iterate on
children of the given missing revisions.

This is a case where staying on the same side of the interface
between the two languages has clear benefits.

On discoveries with initial undecided sets
small enough to bypass sampling entirely, the total cost of
computing the children cache and the subsequent iteration
becomes better than the Python + C counterpart, which relies on
reachableroots2.

For example, on a repo with more than one million revisions with
an initial undecided set of 11 elements, we get these figures:

Rust version with simple iteration
  addcommons: 57.287us
  first undecided computation: 184.278334ms
  first children cache computation: 131.056us
  addmissings iteration: 42.766us
  first addinfo total: 185.24 ms

Python + C version
  first addcommons: 0.29 ms
  addcommons 0.21 ms
  first undecided computation 191.35 ms
  addmissings 45.75 ms
  first addinfo total: 237.77 ms

On discoveries with large undecided sets, the initial price paid
makes the first addinfo slower than the Python + C version,
but that's more than compensated by the gain in sampling and
subsequent iterations.
Here's an extreme example with an undecided set of a million revisions:

Rust version:
  first undecided computation: 293.842629ms
  first children cache computation: 407.911297ms
  addmissings iteration: 34.312869ms
  first addinfo total: 776.02 ms
  taking initial sample
  query 2: sampling time: 1318.38 ms
  query 2; still undecided: 1005013, sample size is: 200
  addmissings: 143.062us

Python + C version:
  first undecided computation 298.13 ms
  addmissings 80.13 ms
  first addinfo total: 399.62 ms
  taking initial sample
  query 2: sampling time: 3957.23 ms
  query 2; still undecided: 1005013, sample size is: 200
  addmissings 52.88 ms

Differential Revision: https://phab.mercurial-scm.org/D6428

== New Features ==

 * New config `commands.commit.post-status` shows status after successful
 commit.

 * `hg root` now has templating support, including support for showing
   where a repo share's source is. See `hg help -v root` for details.

 * New `--force-close-branch` flag for `hg commit` to forcibly close
   branch from a non-head changeset.

 * The curses-based interface for commands like `hg commit -i` now supports
   a range-select mechanism. Select the first line using space like before,
   navigate to the last line, and press X (capital x) to set all items in
   the range at once. Lowercase x has been added as a synonym for space to
   help reinforce the mechanism, and pressing enter/return continues to be a
   synonym for "toggle the current line and move down to the next item in
   this section."

== New Experimental Features ==

 * New config `experimental.log.topo` makes `hg log -G` use
   topological sorting. This is especially useful for aliases since it
   lets the alias accept an `-r` option while still using topological
   sorting with or without the `-r` (unlike if you use the `sort(...,
   topo)` revset).


== Bug Fixes  ==

 * issue4292: "hg log and {files} {file_adds} {file_mods} {file_dels}
   in template show wrong files on merged revision". See details in
   "Backwards Compatibility Changes".


== Backwards Compatibility Changes ==

 * Removed (experimental) support for log graph lines mixing
   parent/grandparent styles. Setting
   e.g. `experimental.graphstyle.parent = !` and
   `experimental.graphstyle.grandparent = 3.` would use `!` for the
   first three lines of the graph and then `.`. This is no longer
   supported.

 * If `ui.origbackuppath` had been (incorrectly) configured to point
   to a file, we will now replace that file by a directory and put
   backups in that directory. This is similar to how we would
   previously replace files *in* the configured directory by
   subdirectories.

* Template keyword `{file_mods}`, `{file_adds}`, and `{file_dels}`
   have changed behavior on merge commits. They used to be relative to
   the first parent, but they now consider both parents. `{file_adds}`
   shows files that exists in the commit but did not exist in either
   parent. `{file_dels}` shows files that do not exist in the commit
   but existed in either parent. `{file_mods}` show the remaining
   files from `{files}` that were not in the other two
   sets.


== Internal API Changes ==

 * Matchers are no longer iterable. Use `match.files()` instead.

 * `match.visitdir()` and `match.visitchildrenset()` now expect the
   empty string instead of '.' to indicate the root directory.

 * `util.dirs()` and `util.finddirs()` now include an entry for the
   root directory (empty string).

 * shelve is no longer an extension now. it will be turned on by default.

 * New API to manage unfinished operations: Earlier there were distinct APIs
   which dealt with unfinished states and separate lists maintaining them
   that are `cmdutil.afterresolvestates`, `cmdutil.unfinishedstates` and
   `cmdutil.STATES`. Now these have been unified to a single
   API which handles the various states and their utilities. This API
   has been added to `state.py`. Now instead of adding to these 3 lists
   independently a state for a new operation can be registered using
   `addunfinished()` in `state` module.

 * `cmdutil.checkunfinished()` now includes detection for merge too.

 * merge abort has been disallowed in case an operation of higher
   precedence is in progress to avoid cases of partial abort of
   operations.

 * We used to automatically attempt to make extensions compatible with
   Python 3 (by translating their source code while loading it). We no
   longer do that.