This is an archive of the discontinued Mercurial Phabricator instance.

Differential D7932

debugbackupbundle: introduce command to interact with strip backups
ClosedPublic

Authored by pulkit on Jan 17 2020, 1:25 PM.

Details

Reviewers
durin42
Group Reviewers
hg-reviewers
Commits
rHGf82d2d4e71db: debugbackupbundle: introduce command to interact with strip backups
rHG89323f85176e: debugbackupbundle: introduce command to interact with strip backups
Summary

This vendors backups extension from hg-experimental.

Listing backups and having some utility to apply them is nice. I know we have
obsmarkers now, but this will help a lot of end users who still uses strip until
we get evolve out of experimental.

Diff Detail

Repository
rHG Mercurial
Branch
default
Lint
No Linters Available
Unit
No Unit Test Coverage

Event Timeline

pulkit created this revision.Jan 17 2020, 1:25 PM
durin42 added a subscriber: durin42.Jan 22 2020, 11:40 AM

I'm +1 on having this functionality in core, FYI.

marmoute added a subscriber: marmoute.Jan 27 2020, 4:59 AM

Can we call it debugbackupbundle for clarify?

pulkit retitled this revision from [RFC]debugbackups: introduce command to interact with strip backups to debugstripbackups: introduce command to interact with strip backups.Feb 4 2020, 8:03 AM
pulkit edited the summary of this revision. (Show Details)
pulkit updated this revision to Diff 19868.
durin42 requested changes to this revision.Feb 12 2020, 2:29 PM

I think I'd prefer debugbackupbundle per marmoute's suggestion.

This revision now requires changes to proceed.Feb 12 2020, 2:29 PM
pulkit retitled this revision from debugstripbackups: introduce command to interact with strip backups to debugbackupbundle: introduce command to interact with strip backups.Feb 24 2020, 1:11 PM
pulkit updated this revision to Diff 20288.
pulkit added a commit: rHG89323f85176e: debugbackupbundle: introduce command to interact with strip backups.Mar 3 2020, 1:48 PM
This revision was not accepted when it landed; it landed in state Needs Review.
Closed by commit rHG89323f85176e: debugbackupbundle: introduce command to interact with strip backups (authored by pulkit). · Explain Why
This revision was automatically updated to reflect the committed changes.
martinvonz added a subscriber: martinvonz.Mar 9 2020, 12:26 PM

I know this has already been queue and I'm just slow, but could you send follow-up patches to address my comments?

mercurial/debugcommands.py
3437

Do we really want to support all of these? (FYI, they include --stat, --graph, --style, --patch, and a few more.)

3453

I think Windows doesn't care much about / vs \, but should we ideally use os.path.join(..., ".hg") here?

3457

Clearer to pass bundlename="" in the call to getremotechanges()?

3458

Clearer to pass force=None (or False?) in the call to getremotechanges()?

3462

Always false, so delete it?

3539–3549

Perhaps the template here should apply at the bundle level instead of the changeset level, so the user can also template the date (and maybe bundle name).

3549

heh, status.modified seems like quite an abuse of the labeling system (I assume it was picked because the author thought that the color they had configured for that looked good in this context too). Could you switch to a different label?

3551

Does False (for the differ argument) produce different output than the default (None) would? If not, remove the False? If it does, could you pass it as a keyword argument instead (differ=False)?

mharbison72 added a subscriber: mharbison72.Mar 10 2020, 3:02 PM
mharbison72 added inline comments.
mercurial/debugcommands.py
3453

I thought we were trying to avoid os.path, so maybe just repo.vfs.join("strip-backup", ".hg")?

(And don't these need to be b'' prefixed for py3?)

marmoute added inline comments.Mar 10 2020, 8:22 PM
mercurial/debugcommands.py
3453

+1 on using vfs here, this is the abstraction we use everywhere else.

Revision Contents
Changeset List

PathPackages
Mmercurial/debugcommands.py (134 lines)
AMtests/debugbackupbundle.t (39 lines)

Diff 20288

mercurial/debugcommands.py

# debugcommands.py - command processing for debug* commands # debugcommands.py - command processing for debug* commands
# #
# Copyright 2005-2016 Matt Mackall <mpm@selenic.com> # Copyright 2005-2016 Matt Mackall <mpm@selenic.com>
# #
# This software may be used and distributed according to the terms of the # This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version. # GNU General Public License version 2 or any later version.
from __future__ import absolute_import from __future__ import absolute_import
import codecs import codecs
import collections import collections
import difflib import difflib
import errno import errno
import glob
import operator import operator
import os import os
import platform import platform
import random import random
import re import re
import socket import socket
import ssl import ssl
import stat import stat
short, short,
) )
from .pycompat import ( from .pycompat import (
getattr, getattr,
open, open,
) )
from . import ( from . import (
bundle2, bundle2,
bundlerepo,
changegroup, changegroup,
cmdutil, cmdutil,
color, color,
context, context,
copies, copies,
dagparser, dagparser,
encoding, encoding,
error, error,
ui.status(_(b'done.\n')) ui.status(_(b'done.\n'))
else: else:
ui.status(_(b'full certificate chain is available\n')) ui.status(_(b'full certificate chain is available\n'))
finally: finally:
s.close() s.close()
@command( @command(
"debugbackupbundle",
[
(
"",
"recover",
"",
"brings the specified changeset back into the repository",
)
]
+ cmdutil.logopts,
martinvonzUnsubmitted
Not Done

Do we really want to support all of these? (FYI, they include --stat, --graph, --style, --patch, and a few more.)

martinvonz: Do we really want to support all of these? (FYI, they include `--stat`, `--graph`, `--style`, `…
_("hg debugbackupbundle [--recover HASH]"),
)
def debugbackupbundle(ui, repo, *pats, **opts):
"""lists the changesets available in backup bundles
Without any arguments, this command prints a list of the changesets in each
backup bundle.
--recover takes a changeset hash and unbundles the first bundle that
contains that hash, which puts that changeset back in your repository.
--verbose will print the entire commit message and the bundle path for that
backup.
"""
backups = filter(
os.path.isfile, glob.glob(repo.vfs.join("strip-backup") + "/*.hg")
martinvonzUnsubmitted
Not Done

I think Windows doesn't care much about / vs \, but should we ideally use os.path.join(..., ".hg") here?

martinvonz: I think Windows doesn't care much about `/` vs `\`, but should we ideally use `os.path.join(...
mharbison72Unsubmitted
Not Done

I thought we were trying to avoid os.path, so maybe just repo.vfs.join("strip-backup", ".hg")?

(And don't these need to be b'' prefixed for py3?)

mharbison72: I thought we were trying to avoid `os.path`, so maybe just `repo.vfs.join("strip-backup", ".
marmouteUnsubmitted
Not Done

+1 on using vfs here, this is the abstraction we use everywhere else.

marmoute: +1 on using `vfs` here, this is the abstraction we use everywhere else.
)
backups.sort(key=lambda x: os.path.getmtime(x), reverse=True)
opts["bundle"] = ""
martinvonzUnsubmitted
Not Done

Clearer to pass bundlename="" in the call to getremotechanges()?

martinvonz: Clearer to pass `bundlename=""` in the call to `getremotechanges()`?
opts["force"] = None
martinvonzUnsubmitted
Not Done

Clearer to pass force=None (or False?) in the call to getremotechanges()?

martinvonz: Clearer to pass `force=None` (or `False`?) in the call to `getremotechanges()`?
limit = logcmdutil.getlimit(opts)
def display(other, chlist, displayer):
if opts.get("newest_first"):
martinvonzUnsubmitted
Not Done

Always false, so delete it?

martinvonz: Always false, so delete it?
chlist.reverse()
count = 0
for n in chlist:
if limit is not None and count >= limit:
break
parents = [True for p in other.changelog.parents(n) if p != nullid]
if opts.get("no_merges") and len(parents) == 2:
continue
count += 1
displayer.show(other[n])
recovernode = opts.get("recover")
if recovernode:
if scmutil.isrevsymbol(repo, recovernode):
ui.warn(_("%s already exists in the repo\n") % recovernode)
return
elif backups:
msg = _(
"Recover changesets using: hg debugbackupbundle --recover "
"<changeset hash>\n\nAvailable backup changesets:"
)
ui.status(msg, label="status.removed")
else:
ui.status(_("no backup changesets found\n"))
return
for backup in backups:
# Much of this is copied from the hg incoming logic
source = ui.expandpath(os.path.relpath(backup, encoding.getcwd()))
source, branches = hg.parseurl(source, opts.get("branch"))
try:
other = hg.peer(repo, opts, source)
except error.LookupError as ex:
msg = _("\nwarning: unable to open bundle %s") % source
hint = _("\n(missing parent rev %s)\n") % short(ex.name)
ui.warn(msg, hint=hint)
continue
revs, checkout = hg.addbranchrevs(
repo, other, branches, opts.get("rev")
)
if revs:
revs = [other.lookup(rev) for rev in revs]
quiet = ui.quiet
try:
ui.quiet = True
other, chlist, cleanupfn = bundlerepo.getremotechanges(
ui, repo, other, revs, opts["bundle"], opts["force"]
)
except error.LookupError:
continue
finally:
ui.quiet = quiet
try:
if not chlist:
continue
if recovernode:
with repo.lock(), repo.transaction("unbundle") as tr:
if scmutil.isrevsymbol(other, recovernode):
ui.status(_("Unbundling %s\n") % (recovernode))
f = hg.openpath(ui, source)
gen = exchange.readbundle(ui, f, source)
if isinstance(gen, bundle2.unbundle20):
bundle2.applybundle(
repo,
gen,
tr,
source="unbundle",
url="bundle:" + source,
)
else:
gen.apply(repo, "unbundle", "bundle:" + source)
break
else:
backupdate = time.strftime(
"%a %H:%M, %Y-%m-%d",
time.localtime(os.path.getmtime(source)),
)
ui.status("\n%s\n" % (backupdate.ljust(50)))
if ui.verbose:
ui.status("%s%s\n" % ("bundle:".ljust(13), source))
else:
opts[
"template"
] = "{label('status.modified', node|short)} {desc|firstline}\n"
martinvonzUnsubmitted
Not Done

heh, status.modified seems like quite an abuse of the labeling system (I assume it was picked because the author thought that the color they had configured for that looked good in this context too). Could you switch to a different label?

martinvonz: heh, `status.modified` seems like quite an abuse of the labeling system (I assume it was picked…
martinvonzUnsubmitted
Not Done

Perhaps the template here should apply at the bundle level instead of the changeset level, so the user can also template the date (and maybe bundle name).

martinvonz: Perhaps the template here should apply at the bundle level instead of the changeset level, so…
displayer = logcmdutil.changesetdisplayer(
ui, other, opts, False
martinvonzUnsubmitted
Not Done

Does False (for the differ argument) produce different output than the default (None) would? If not, remove the False? If it does, could you pass it as a keyword argument instead (differ=False)?

martinvonz: Does `False` (for the `differ` argument) produce different output than the default (`None`)…
)
display(other, chlist, displayer)
displayer.close()
finally:
cleanupfn()
@command(
b'debugsub', b'debugsub',
[(b'r', b'rev', b'', _(b'revision to check'), _(b'REV'))], [(b'r', b'rev', b'', _(b'revision to check'), _(b'REV'))],
_(b'[-r REV] [REV]'), _(b'[-r REV] [REV]'),
) )
def debugsub(ui, repo, rev=None): def debugsub(ui, repo, rev=None):
ctx = scmutil.revsingle(repo, rev, None) ctx = scmutil.revsingle(repo, rev, None)
for k, v in sorted(ctx.substate.items()): for k, v in sorted(ctx.substate.items()):
ui.writenoi18n(b'path %s\n' % k) ui.writenoi18n(b'path %s\n' % k)

tests/debugbackupbundle.t

  • This file was added.
$ cat >> $HGRCPATH << EOF
> [extensions]
> strip=
> EOF
Setup repo
$ hg init repo
$ cd repo
Test backups list and recover
$ hg debugbackupbundle
no backup changesets found
$ mkcommit() {
> echo "$1" > "$1"
> hg add "$1"
> hg ci -l $1
> }
$ mkcommit a
$ mkcommit b
$ hg strip .
0 files updated, 0 files merged, 1 files removed, 0 files unresolved
saved backup bundle to $TESTTMP/repo/.hg/strip-backup/d2ae7f538514-2953539b-backup.hg (glob)
$ hg debugbackupbundle
Recover changesets using: hg debugbackupbundle --recover <changeset hash>
Available backup changesets:
* (glob)
d2ae7f538514 b
$ hg debugbackupbundle --recover d2ae7f538514
Unbundling d2ae7f538514
adding changesets
adding manifests
adding file changes
added 1 changesets with 1 changes to 1 files
new changesets d2ae7f538514 (1 drafts)