It gives me great pleasure to release another, yet-more-improved version
of the Perl history.
I triaged and dug to the bottom of each of the faults that Nicholas
raised last month. Using his QA approach as a guideline, I was able to
use the perforce exporter I wrote for this effort repeatedly until I
ended up with output which is largely identical to the perforce source.
After a brief discussion on #p5p, it was decided that rather than
messing around with history patches that would need to be grafted on by
those who wanted the cleaned history for bisecting etc, that it would be
easier to just force everyone to re-clone and have one good history.
The new version is available at:
git://utsl.gen.nz/perl-sotomorrow
It's now about an 75MB download (37k revisions). Much reduced from
previous downloads, without investigating I'd say that boils down to
improvements to the git deltification and object candidate selection
code, resulting in more, longer string matches. Or, maybe I just left a
big chunk of important stuff out, who knows ;-).
I'll keep it a bit more up to date than the other one has been
historically - assuming my upstream also updates beyond 13th November
soon.
And for those interested, here is a blow-by-blow account of each of the
issues raised, and what happened with them:
* Change 42 (maint-5.004/perl):
- Change 42 was the commit that introduced perl-5.004_01 to
Perforce. Its parent was Change 32. However, there was a fine
grained changelog for this, so it was expanded to approximately 47
commits using the "Timinator" script.
=> There was a tag in the 'soyesterday' release of the Perl history
so that 'git log --all --grep=@42\\\>' will show up a revision with
@42 on it.
* Change 82 (maint-5.004/perl):
- Change 82 was where the 5.004_03 revision was checked into
Perforce. The Timinator extracted 29 commits from that change; it
was recorded as a cross-merge from the surrogate Change 81 and the
actual commit tagged with perl-5.004_03.
=> Actually the error comes from change 32, which because I was kind
of manually doing those maint-5.004 revisions using the
timinator, didn't get the attention required.
So, when I rewound the exporter to that change and re-exported
it, it comes up clean:
maia:~/src/pumpkin/perl$ git-p4raw unexport-commits 32
git-p4raw: deleting commit records for 34714 changes
maia:~/src/pumpkin/perl$ rm -r .git/refs/heads/p4/
maia:~/src/pumpkin/perl$ git-p4raw show-branches 31 | \
sed 's!//depot!refs/heads/p4!' | \
while read ref x commit; do git-update-ref $ref $commit; done
maia:~/src/pumpkin/perl$ git-p4raw export-commits -n 1
git-p4raw: gathering export plan
git-p4raw: exporting commits between 32 and 32
100%
[=================================================================================]D 0h00m00sgit-p4raw: Now checkpointing.
git-p4raw: waited 1s for p4raw.29232.marks to be created
maia:~/src/pumpkin/perl$ git checkout p4/maint-5.004/perl
Switched to branch "p4/maint-5.004/perl"
maia:~/src/pumpkin/perl$ diff -rpu . ../p4perl/maint-5.004/perl
Only in .: .git
Only in .: p4raw.29232.marks
maia:~/src/pumpkin/perl$
* Change 84 (blead):
- Change 84 was a cross-merge from 5.004_01 (Change 42) and blead
(Change 78)
=> diffs seemed to come out in the wash.
* Change 157 (oneperl):
- See the note on change 162. This manual fixup was possibly
botched previously; it's clean now, though.
* Change 192 (oneperl):
=> related to other oneperl issues
* Change 562 (blead):
- Changes 562 through 564 were quite special, a whole set of partial
merges of the p4/perlext/Compiler tree, with re-organisation. In
the history, p4/perlext/Compiler appears as a new root and is
merged in at change 563.
Indeed the export of Change 562 was completely screwed.
=> rotated change several turns anti-clockwise, now completely
un-screwed
* Change 973 (win32/perl):
- Another "nuke everything" commit. There was a lot of manual work
around that area, and the old buggy importer probably didn't help
matters.
=> all the non-'fixed' history has been checked, double checked and
re-written.
Change 986 is the end of the "manual" history - I didn't bother
running the timinator over later changes on the maint-5.004
branch. That history - particularly commit cb99a88 - has been
left untouched, as the only problems are a few e-mail addresses
and some of the 'p4raw-link' links will refer to unknown revisions
- and fixing that up would be a bit of a headache right now.
* Change 2041 (perlext/jpl):
- Fixed symlink handling, this is the first change with a symlink.
Affects many revisions.
* Change 4249 (vmsperl):
- Export of this revision only affected one file.
missing: configure.com vms/vms.c
also missing (may be related to Change 82):
ext/Thread/join.t ext/Thread/specific.t lib/warning.pm
op/delete.t op/flip.t op/push.t op/wantarray.t warning.h
x2p/str.c
=> another one fixed in the wash.
* Change 8520:
- This included a file with keyword expansion. In theory this work
directory corruption can be achieved via gitattributes; but, I
didn't bother; those features never really work ;-).
=> WONTFIX
* Change 11243:
- Change 11243 contained files which had a file type of XXX - "mac"
Sorry, I didn't do any .gitattributes conversion for these either.
It would probably need more infrastructure - currently the
'smudge' and 'clean' filters are just that - stream filters - and
there is no infrastructure for unpacking single repository files
to multiple working copy files (or vice versa, a possible useful
feature for storing collections of, eg tar archives).
=> WONTFIX
* Change 16123:
- This is a pretty kooky change. It created a new branch, and
pulled in a tree from another branch. It was represented in Git
as a cross-merge between the two sources of changes.
However, this strange mish-mash of a situation represents a novel
condition to the importer; it needed to spot that the path doesn't
match the old path and issue a "deleteall" git-fast-import
instruction to clear the branch.
=> implemented, and works.
* Change 29366:
- This change added a file with a literal "#" in the filename. This
is indeed a deficiency in my perforce converter; that's the first
file named like that I found. Fixed the conversion and now the
short-lived file appears and disappears correctly named.
Other things not fixed and general errata:
------------------------------------------
- The maint-5.004 track has some dead 'p4raw-link' comments; I didn't
change these as mentioned earlier. Complicated rewrites to fix them
up could have made matters worse.
- The maint-5.004 and one other branch has some files which are
execute enabled in Perforce but not in the git version; this was a
result of the manual conversion work for the earlier history.
- One of the manual corrections I made around change 136 - 151 was
missed, so the history for those changes on the affected branch
(probably oneperl) may be a little screwy.
- The amend note tags present in the perl-soyesterday repository have
not been migrated to the final repository; I'll leave them in
perl-soyesterday.
Enjoy!
Sam.
|
Highlight