This is an archive of the discontinued Mercurial Phabricator instance.

Differential D876

remotefilelog: update to new deltachunk return value
ClosedPublic

Authored by ryanmce on Oct 1 2017, 7:54 AM.
Tags
None
Subscribers
mbthomas

Details

Reviewers
mbthomas
Group Reviewers
Restricted Project
Commits
rFBHGX96d881ce2cf4: remotefilelog: update to new deltachunk return value
Summary

In D746, we changed the return value of deltachunk. This diff makes me
a bit sad but it fixes remotefilelog to work with those changes.

Test Plan

tests

Diff Detail

Repository
rFBHGX Facebook Mercurial Extensions
Branch
default
Lint
Lint OK
Unit
Unit Tests OK

Event Timeline

ryanmce created this revision.Oct 1 2017, 7:54 AM
ryanmce updated this revision to Diff 2275.Oct 1 2017, 9:02 AM

rerun atop a fixed up stack

ryanmce updated this revision to Diff 2279.Oct 1 2017, 9:07 AM

reverse order of commits

mbthomas accepted this revision.Oct 1 2017, 10:02 AM
mbthomas added a subscriber: mbthomas.
mbthomas added inline comments.
remotefilelog/shallowbundle.py
363

I don't really like indexed lookup of tuples (prefer unpacking), but this is probably ok as we only need a few things.

This revision is now accepted and ready to land.Oct 1 2017, 10:02 AM
ryanmce added a comment.Oct 1 2017, 10:12 AM

This is also why this diff makes me sad. Looks like a readability loss to me.

remotefilelog/shallowbundle.py
363

This is why a style czar should declare that "records are the one true way". Alas, this is what we get for now.

Closed by commit rFBHGX96d881ce2cf4: remotefilelog: update to new deltachunk return value (authored by ryanmce). · Explain WhyOct 1 2017, 10:13 AM
This revision was automatically updated to reflect the committed changes.

Revision Contents
Changeset List

PathPackages
Mremotefilelog/shallowbundle.py (14 lines)

Diff 2279

remotefilelog/shallowbundle.py

if not repo.shallowmatch(f): if not repo.shallowmatch(f):
fl = repo.file(f) fl = repo.file(f)
deltas = source.deltaiter() deltas = source.deltaiter()
fl.addgroup(deltas, revmap, trp) fl.addgroup(deltas, revmap, trp)
continue continue
chain = None chain = None
while True: while True:
# returns: (node, p1, p2, cs, deltabase, delta, flags) or None
revisiondata = source.deltachunk(chain) revisiondata = source.deltachunk(chain)
if not revisiondata: if not revisiondata:
break break
chain = revisiondata['node'] chain = revisiondata[0]
revisiondatas[(f, chain)] = revisiondata revisiondatas[(f, chain)] = revisiondata
queue.append((f, chain)) queue.append((f, chain))
if f not in visited: if f not in visited:
newfiles += 1 newfiles += 1
visited.add(f) visited.add(f)
return True return True
skipcount = 0 skipcount = 0
# Prefetch the non-bundled revisions that we will need # Prefetch the non-bundled revisions that we will need
prefetchfiles = [] prefetchfiles = []
for f, node in queue: for f, node in queue:
revisiondata = revisiondatas[(f, node)] revisiondata = revisiondatas[(f, node)]
dependents = [revisiondata['p1'], revisiondata['p2'], # revisiondata: (node, p1, p2, cs, deltabase, delta, flags)
revisiondata['deltabase']] dependents = [revisiondata[1], revisiondata[2], revisiondata[4]]
mbthomasUnsubmitted
Not Done

I don't really like indexed lookup of tuples (prefer unpacking), but this is probably ok as we only need a few things.

mbthomas: I don't really like indexed lookup of tuples (prefer unpacking), but this is probably ok as we…
ryanmceAuthorUnsubmitted
Not Done

This is why a style czar should declare that "records are the one true way". Alas, this is what we get for now.

ryanmce: This is why a style czar should declare that "records are the one true way". Alas, this is what…
for dependent in dependents: for dependent in dependents:
if dependent == nullid or (f, dependent) in revisiondatas: if dependent == nullid or (f, dependent) in revisiondatas:
continue continue
prefetchfiles.append((f, hex(dependent))) prefetchfiles.append((f, hex(dependent)))
repo.fileservice.prefetch(prefetchfiles) repo.fileservice.prefetch(prefetchfiles)
# Apply the revisions in topological order such that a revision # Apply the revisions in topological order such that a revision
# is only written once it's deltabase and parents have been written. # is only written once it's deltabase and parents have been written.
while queue: while queue:
f, node = queue.pop(0) f, node = queue.pop(0)
if (f, node) in processed: if (f, node) in processed:
continue continue
skipcount += 1 skipcount += 1
if skipcount > len(queue) + 1: if skipcount > len(queue) + 1:
raise error.Abort(_("circular node dependency")) raise error.Abort(_("circular node dependency"))
fl = repo.file(f) fl = repo.file(f)
revisiondata = revisiondatas[(f, node)] revisiondata = revisiondatas[(f, node)]
p1 = revisiondata['p1'] # revisiondata: (node, p1, p2, cs, deltabase, delta, flags)
p2 = revisiondata['p2'] node, p1, p2, linknode, deltabase, delta, flags = revisiondata
linknode = revisiondata['cs']
deltabase = revisiondata['deltabase']
delta = revisiondata['delta']
if not available(f, node, f, deltabase): if not available(f, node, f, deltabase):
continue continue
base = fl.revision(deltabase, raw=True) base = fl.revision(deltabase, raw=True)
text = mdiff.patch(base, delta) text = mdiff.patch(base, delta)
if isinstance(text, buffer): if isinstance(text, buffer):
text = str(text) text = str(text)