Patchwork D134: rebase: add config to move rebase into a single transaction

login
register
mail settings
Submitter phabricator
Date July 18, 2017, 3:09 p.m.
Message ID <differential-rev-PHID-DREV-l57rdikzv6w5r72mpc2d-req@phab.mercurial-scm.org>
Download mbox | patch
Permalink /patch/22492/
State Superseded, archived
Headers show

Comments

phabricator - July 18, 2017, 3:09 p.m.
durham created this revision.
Herald added a subscriber: mercurial-devel.
Herald added a reviewer: hg-reviewers.

REVISION SUMMARY
  This was previously landed as https://phab.mercurial-scm.org/rHGcf8ad0e6c0e4555809c8e7232c8819ecf152bf1b but backed out in https://phab.mercurial-scm.org/rHGa5abaa81fad665602821b939ca4101d48f4d0ef7 because
  it broke hook mid rebase and caused conflict resolution data loss in the event
  of unexpected exceptions. This new version adds the behavior back but behind a
  config flag, since the performance improvement is notable in large repositories.
  
  The next patch adds a test covering this config.
  
  The old commit message was:
  
  Previously, rebasing would open several transaction over the course of rebasing
  several commits. Opening a transaction can have notable overhead (like copying
  the dirstate) which can add up when rebasing many commits.
  
  This patch adds a single large transaction around the actual commit rebase
  operation, with a catch for intervention which serializes the current state if
  we need to drop back to the terminal for user intervention. Amazingly, almost
  all the tests seem to pass.
  
  On large repos with large working copies, this can speed up rebasing 7 commits
  by 25%. I'd expect the percentage to be a bit larger for rebasing even more
  commits.
  
  There are minor test changes because we're rolling back the entire transaction
  during unexpected exceptions instead of just stopping mid-rebase, so there's no
  more backup bundle. It also leave an unknown file in the working copy, since our
  clean up 'hg update' doesn't delete unknown files.
  (grafted from cca36c7f35261b0e31beb226bf361067ef0e06ab)
  (grafted from dc497d8705b71503e32e07bd33925c1e42cf9c9a)

REPOSITORY
  rHG Mercurial

REVISION DETAIL
  https://phab.mercurial-scm.org/D134

AFFECTED FILES
  hgext/rebase.py
  mercurial/util.py

CHANGE DETAILS




EMAIL PREFERENCES
  https://phab.mercurial-scm.org/settings/panel/emailpreferences/

To: durham, #hg-reviewers
Cc: mercurial-devel
phabricator - July 18, 2017, 3:29 p.m.
martinvonz added inline comments.

INLINE COMMENTS

> rebase.py:712-718
> +        singletr = ui.configbool('rebase', 'singletransaction')
> +        if singletr:
> +            tr = repo.transaction('rebase')
> +        else:
> +            tr = util.nullcontextmanager()
> +        with util.acceptintervention(tr):
> +            rbsrt._performrebase(tr if singletr else None)

acceptintervention() also accepts None (it's a little unexpected, I know, but also convenient), so this can probably be:

  tr = None
  if singletr:
      tr = repo.transaction('rebase')  
  with util.acceptintervention(tr):
      rbsrt._performrebase(tr)

Also get rid of nullcontextmanager() then of course

REPOSITORY
  rHG Mercurial

REVISION DETAIL
  https://phab.mercurial-scm.org/D134

EMAIL PREFERENCES
  https://phab.mercurial-scm.org/settings/panel/emailpreferences/

To: durham, #hg-reviewers
Cc: martinvonz, mercurial-devel

Patch

diff --git a/mercurial/util.py b/mercurial/util.py
--- a/mercurial/util.py
+++ b/mercurial/util.py
@@ -602,6 +602,20 @@ 
     finally:
         tr.release()
 
+class nullcontextmanager(object):
+    """A no-op context manager.
+    """
+    def __enter__(self):
+        return self
+    def __exit__(self, exc_type, exc_val, exc_tb):
+        pass
+    def close(self):
+        pass
+    def release(self):
+        pass
+    def abort(self):
+        pass
+
 class _lrucachenode(object):
     """A node in a doubly linked list.
 
diff --git a/hgext/rebase.py b/hgext/rebase.py
--- a/hgext/rebase.py
+++ b/hgext/rebase.py
@@ -343,7 +343,7 @@ 
         if dest.closesbranch() and not self.keepbranchesf:
             self.ui.status(_('reopening closed branch head %s\n') % dest)
 
-    def _performrebase(self):
+    def _performrebase(self, tr):
         repo, ui, opts = self.repo, self.ui, self.opts
         if self.keepbranchesf:
             # insert _savebranch at the start of extrafns so if
@@ -394,7 +394,7 @@ 
                                              self.state,
                                              self.destancestors,
                                              self.obsoletenotrebased)
-                self.storestatus()
+                self.storestatus(tr=tr)
                 storecollapsemsg(repo, self.collapsemsg)
                 if len(repo[None].parents()) == 2:
                     repo.ui.debug('resuming interrupted rebase\n')
@@ -641,6 +641,15 @@ 
       [commands]
       rebase.requiredest = True
 
+    By default, rebase will close the transaction after each commit. For
+    performance purposes, you can configure rebase to use a single transaction
+    across the entire rebase. WARNING: This setting introduces a significant
+    risk of losing the work you've done in a rebase if the rebase aborts
+    unexpectedly::
+
+      [rebase]
+      singletransaction = True
+
     Return Values:
 
     Returns 0 on success, 1 if nothing to rebase or there are
@@ -700,7 +709,14 @@ 
             if retcode is not None:
                 return retcode
 
-        rbsrt._performrebase()
+        singletr = ui.configbool('rebase', 'singletransaction')
+        if singletr:
+            tr = repo.transaction('rebase')
+        else:
+            tr = util.nullcontextmanager()
+        with util.acceptintervention(tr):
+            rbsrt._performrebase(tr if singletr else None)
+
         rbsrt._finishrebase()
 
 def _definesets(ui, repo, destf=None, srcf=None, basef=None, revf=None,