This is an archive of the discontinued Mercurial Phabricator instance.

Differential D3873

rebase: refactor logic to read rebasestate in a separate function
ClosedPublic

Authored by pulkit on Jul 1 2018, 4:32 PM.

Details

Reviewers
None
Group Reviewers
hg-reviewers
Commits
rHG19076e2d62e7: rebase: refactor logic to read rebasestate in a separate function
Summary

This will help us in plugging the use of state.cmdstate() to read rebasestate.

Diff Detail

Repository
rHG Mercurial
Lint
Lint Skipped
Unit
Unit Tests Skipped

Event Timeline

pulkit created this revision.Jul 1 2018, 4:32 PM
Closed by commit rHG19076e2d62e7: rebase: refactor logic to read rebasestate in a separate function (authored by pulkit). · Explain WhyJul 2 2018, 8:43 AM
This revision was automatically updated to reflect the committed changes.
yuja added a subscriber: yuja.Jul 2 2018, 8:45 AM

Queued, thanks.

f = repo.vfs("rebasestate")
for i, l in enumerate(f.read().splitlines()):
    if i == 0:
  • originalwd = repo[l].rev()

+ data['originalwd'] = repo[l].rev()

FWIW, I don't have any better idea to keep old hg not crash with new state
file, other than:

a. use separate file (e.g. "rebasestate2") and leave "rebasestate" in old

format (or make it an empty file to trigger error.)

b. abuse repo[l].rev() to abort with lookup error with somewhat descriptive

message.
pulkit added a comment.Jul 2 2018, 2:18 PM
In D3873#60501, @yuja wrote:

Queued, thanks.

f = repo.vfs("rebasestate")
for i, l in enumerate(f.read().splitlines()):
    if i == 0:
  • originalwd = repo[l].rev()

+ data['originalwd'] = repo[l].rev()

FWIW, I don't have any better idea to keep old hg not crash with new state
file, other than:
a. use separate file (e.g. "rebasestate2") and leave "rebasestate" in old

format (or make it an empty file to trigger error.)

Yeah, I was preventing this for a long time but feels like I can't anymore. How about having a .hg/state/ where we will have our new histedit and rebase state files?

b. abuse repo[l].rev() to abort with lookup error with somewhat descriptive

message.

This is more bad. Let's not do that.

yuja added a comment.Jul 3 2018, 8:02 AM
> FWIW, I don't have any better idea to keep old hg not crash with new state
>  file, other than:
>
>   a. use separate file (e.g. "rebasestate2") and leave "rebasestate" in old
>      format (or make it an empty file to trigger error.)
Yeah, I was preventing this for a long time but feels like I can't anymore.
How about having a `.hg/state/` where we will have our new histedit and
rebase state files?

A separate directory wouldn't make the situation better. We can't stop writing
a v1 state file anyway since missing v1 state file means "no rebase in progress"
in old hg versions.

Revision Contents
Changeset List

PathPackages
Mhgext/rebase.py (50 lines)

Diff 9395

hgext/rebase.py

else: else:
newrev = "%d" % v newrev = "%d" % v
destnode = repo[destmap[d]].hex() destnode = repo[destmap[d]].hex()
f.write("%s:%s:%s\n" % (oldrev, newrev, destnode)) f.write("%s:%s:%s\n" % (oldrev, newrev, destnode))
repo.ui.debug('rebase status stored\n') repo.ui.debug('rebase status stored\n')
def restorestatus(self): def restorestatus(self):
"""Restore a previously stored status""" """Restore a previously stored status"""
data = self._read()
self.repo.ui.debug('rebase status resumed\n')
self.originalwd = data['originalwd']
self.destmap = data['destmap']
self.state = data['state']
self.skipped = data['skipped']
self.collapsef = data['collapse']
self.keepf = data['keep']
self.keepbranchesf = data['keepbranches']
self.external = data['external']
self.activebookmark = data['activebookmark']
def _read(self):
self.prepared = True self.prepared = True
repo = self.repo repo = self.repo
assert repo.filtername is None assert repo.filtername is None
keepbranches = None data = {'keepbranches': None, 'collapse': None, 'activebookmark': None,
'external': nullrev, 'keep': None, 'originalwd': None}
legacydest = None legacydest = None
collapse = False
external = nullrev
activebookmark = None
state = {} state = {}
destmap = {} destmap = {}
try: try:
f = repo.vfs("rebasestate") f = repo.vfs("rebasestate")
for i, l in enumerate(f.read().splitlines()): for i, l in enumerate(f.read().splitlines()):
if i == 0: if i == 0:
originalwd = repo[l].rev() data['originalwd'] = repo[l].rev()
elif i == 1: elif i == 1:
# this line should be empty in newer version. but legacy # this line should be empty in newer version. but legacy
# clients may still use it # clients may still use it
if l: if l:
legacydest = repo[l].rev() legacydest = repo[l].rev()
elif i == 2: elif i == 2:
external = repo[l].rev() data['external'] = repo[l].rev()
elif i == 3: elif i == 3:
collapse = bool(int(l)) data['collapse'] = bool(int(l))
elif i == 4: elif i == 4:
keep = bool(int(l)) data['keep'] = bool(int(l))
elif i == 5: elif i == 5:
keepbranches = bool(int(l)) data['keepbranches'] = bool(int(l))
elif i == 6 and not (len(l) == 81 and ':' in l): elif i == 6 and not (len(l) == 81 and ':' in l):
# line 6 is a recent addition, so for backwards # line 6 is a recent addition, so for backwards
# compatibility check that the line doesn't look like the # compatibility check that the line doesn't look like the
# oldrev:newrev lines # oldrev:newrev lines
activebookmark = l data['activebookmark'] = l
else: else:
args = l.split(':') args = l.split(':')
oldrev = repo[args[0]].rev() oldrev = repo[args[0]].rev()
newrev = args[1] newrev = args[1]
if newrev in legacystates: if newrev in legacystates:
continue continue
if len(args) > 2: if len(args) > 2:
destrev = repo[args[2]].rev() destrev = repo[args[2]].rev()
else: else:
destrev = legacydest destrev = legacydest
destmap[oldrev] = destrev destmap[oldrev] = destrev
if newrev == revtodostr: if newrev == revtodostr:
state[oldrev] = revtodo state[oldrev] = revtodo
# Legacy compat special case # Legacy compat special case
else: else:
state[oldrev] = repo[newrev].rev() state[oldrev] = repo[newrev].rev()
except IOError as err: except IOError as err:
if err.errno != errno.ENOENT: if err.errno != errno.ENOENT:
raise raise
cmdutil.wrongtooltocontinue(repo, _('rebase')) cmdutil.wrongtooltocontinue(repo, _('rebase'))
if keepbranches is None: if data['keepbranches'] is None:
raise error.Abort(_('.hg/rebasestate is incomplete')) raise error.Abort(_('.hg/rebasestate is incomplete'))
data['destmap'] = destmap
data['state'] = state
skipped = set() skipped = set()
# recompute the set of skipped revs # recompute the set of skipped revs
if not collapse: if not data['collapse']:
seen = set(destmap.values()) seen = set(destmap.values())
for old, new in sorted(state.items()): for old, new in sorted(state.items()):
if new != revtodo and new in seen: if new != revtodo and new in seen:
skipped.add(old) skipped.add(old)
seen.add(new) seen.add(new)
data['skipped'] = skipped
repo.ui.debug('computed skipped revs: %s\n' % repo.ui.debug('computed skipped revs: %s\n' %
(' '.join('%d' % r for r in sorted(skipped)) or '')) (' '.join('%d' % r for r in sorted(skipped)) or ''))
repo.ui.debug('rebase status resumed\n')
self.originalwd = originalwd return data
self.destmap = destmap
self.state = state
self.skipped = skipped
self.collapsef = collapse
self.keepf = keep
self.keepbranchesf = keepbranches
self.external = external
self.activebookmark = activebookmark
def _handleskippingobsolete(self, obsoleterevs, destmap): def _handleskippingobsolete(self, obsoleterevs, destmap):
"""Compute structures necessary for skipping obsolete revisions """Compute structures necessary for skipping obsolete revisions
obsoleterevs: iterable of all obsolete revisions in rebaseset obsoleterevs: iterable of all obsolete revisions in rebaseset
destmap: {srcrev: destrev} destination revisions destmap: {srcrev: destrev} destination revisions
""" """
self.obsoletenotrebased = {} self.obsoletenotrebased = {}