Mercurial > public > mercurial-scm > hg
comparison mercurial/copies.py @ 26013:38f92d12357c
copy: add flag for disabling copy tracing
Copy tracing can be up to 80% of rebase time when rebasing stacks of commits in
large repos (hundreds of thousands of files). This provides the option of
turning off the majority of copy tracing. It does not turn off _forwardcopies()
since that is used to carry copy information inside a commit across a rebase.
This will affect the situation where a user edits a file, then rebases on top of
commits that have moved that file. The move will not be detected and the user
will have to manually resolve the issue (possibly by redoing the rebase with
this flag off).
The reason to have a flag instead of trying to fix the actual copy tracing
performance is that copy tracing is fundamentally an O(number of files in the
repo) operation. In order to know if file X in the rebase source was copied
anywhere, we have to walk the filelog for every new file that exists in the
rebase destination (i.e. a file in the destination that is not in the common
ancestor). Without an index that lets us trace forward (i.e. from file Y in the
common ancestor forward to the rebase destination), it will never be an O(number
of changes in my branch) operation.
In mozilla-central, rebasing a 3 commit stack across 20,000 revs goes from 39s
to 11s.
| author | Durham Goode <durham@fb.com> |
|---|---|
| date | Tue, 27 Jan 2015 11:26:27 -0800 |
| parents | cfc24c22454e |
| children | 3c6902ed9f07 |
comparison
equal
deleted
inserted
replaced
| 26011:ce77436162a5 | 26013:38f92d12357c |
|---|---|
| 189 cm = _chain(a, w, cm, _dirstatecopies(w)) | 189 cm = _chain(a, w, cm, _dirstatecopies(w)) |
| 190 | 190 |
| 191 return cm | 191 return cm |
| 192 | 192 |
| 193 def _backwardrenames(a, b): | 193 def _backwardrenames(a, b): |
| 194 if a._repo.ui.configbool('experimental', 'disablecopytrace'): | |
| 195 return {} | |
| 196 | |
| 194 # Even though we're not taking copies into account, 1:n rename situations | 197 # Even though we're not taking copies into account, 1:n rename situations |
| 195 # can still exist (e.g. hg cp a b; hg mv a c). In those cases we | 198 # can still exist (e.g. hg cp a b; hg mv a c). In those cases we |
| 196 # arbitrarily pick one of the renames. | 199 # arbitrarily pick one of the renames. |
| 197 f = _forwardcopies(b, a) | 200 f = _forwardcopies(b, a) |
| 198 r = {} | 201 r = {} |
| 261 return {}, {}, {}, {} | 264 return {}, {}, {}, {} |
| 262 | 265 |
| 263 # avoid silly behavior for parent -> working dir | 266 # avoid silly behavior for parent -> working dir |
| 264 if c2.node() is None and c1.node() == repo.dirstate.p1(): | 267 if c2.node() is None and c1.node() == repo.dirstate.p1(): |
| 265 return repo.dirstate.copies(), {}, {}, {} | 268 return repo.dirstate.copies(), {}, {}, {} |
| 269 | |
| 270 # Copy trace disabling is explicitly below the node == p1 logic above | |
| 271 # because the logic above is required for a simple copy to be kept across a | |
| 272 # rebase. | |
| 273 if repo.ui.configbool('experimental', 'disablecopytrace'): | |
| 274 return {}, {}, {}, {} | |
| 266 | 275 |
| 267 limit = _findlimit(repo, c1.rev(), c2.rev()) | 276 limit = _findlimit(repo, c1.rev(), c2.rev()) |
| 268 if limit is None: | 277 if limit is None: |
| 269 # no common ancestor, no copies | 278 # no common ancestor, no copies |
| 270 return {}, {}, {}, {} | 279 return {}, {}, {}, {} |
| 511 filter copy records. Any copies that occur between fromrev and | 520 filter copy records. Any copies that occur between fromrev and |
| 512 skiprev will not be duplicated, even if they appear in the set of | 521 skiprev will not be duplicated, even if they appear in the set of |
| 513 copies between fromrev and rev. | 522 copies between fromrev and rev. |
| 514 ''' | 523 ''' |
| 515 exclude = {} | 524 exclude = {} |
| 516 if skiprev is not None: | 525 if (skiprev is not None and |
| 526 not repo.ui.configbool('experimental', 'disablecopytrace')): | |
| 527 # disablecopytrace skips this line, but not the entire function because | |
| 528 # the line below is O(size of the repo) during a rebase, while the rest | |
| 529 # of the function is much faster (and is required for carrying copy | |
| 530 # metadata across the rebase anyway). | |
| 517 exclude = pathcopies(repo[fromrev], repo[skiprev]) | 531 exclude = pathcopies(repo[fromrev], repo[skiprev]) |
| 518 for dst, src in pathcopies(repo[fromrev], repo[rev]).iteritems(): | 532 for dst, src in pathcopies(repo[fromrev], repo[rev]).iteritems(): |
| 519 # copies.pathcopies returns backward renames, so dst might not | 533 # copies.pathcopies returns backward renames, so dst might not |
| 520 # actually be in the dirstate | 534 # actually be in the dirstate |
| 521 if dst in exclude: | 535 if dst in exclude: |