facebook/sapling
1
1
Fork 0
mirror of https://github.com/facebook/sapling.git synced 2024-10-09 00:14:35 +03:00
Code Issues Projects Releases Wiki Activity
8,202 Commits 24 Branches 62 Tags 322 MiB
2c8901a1b9
Commit Graph

254 Commits

Author SHA1 Message Date
Patrick Mezard
327e3eb360 revlog: fix inlined revision transaction extra data (issue 749) 2007-09-23 18:24:19 +02:00
Matt Mackall
ccc415d026 revlog: more robust for damaged indexes 
- fix flag reporting
- make broken length reporting smarter
 2007-09-21 17:14:39 -05:00
Matt Mackall
37cc19c8c5 revlog: localize some fastpath functions 2007-07-26 12:02:58 -05:00
Matt Mackall
eeb8f14a99 revlog: minor chunk speed-up 2007-07-26 12:02:58 -05:00
Matt Mackall
d106fa8150 revlog: minor revdiff reorganization 2007-07-26 12:02:58 -05:00
Matt Mackall
9429a0a390 revlog: move flag checking out of the offset fastpath 2007-07-26 12:02:58 -05:00
Matt Mackall
767bb11dd5 revlog: avoid some unnecessary seek/tell syscalls 2007-07-23 20:44:08 -05:00
Matt Mackall
dd01d60440 lazyparser.findnode: fix typo and s/rfind/find/ 
There's no reason to use reverse string search and it's slightly slower
 2007-07-23 20:44:08 -05:00
Matt Mackall
7292978647 lazyindex: speed up __getitem__ 
This function is fairly performance sensitive, so we make a couple
ugly tweaks:

- keep all entries packed so we needn't test entry types
- fold index lookup/load into unpack call to eliminate
  local variable setting
 2007-07-23 20:44:08 -05:00
Matt Mackall
a68e049384 lazyparser: up the blocksize from 512 bytes to 64k 2007-07-23 20:44:08 -05:00
Matt Mackall
b607d22a05 revlog: implement a fast path for heads 2007-07-23 20:44:08 -05:00
Matt Mackall
cb95c386e8 revlogio: speed up parsing 
- precalcuate ending offset
- pull some variables into local scope
- separate inline and out of line code paths
 2007-07-23 20:44:08 -05:00
Matt Mackall
8fa2498ebf revlog: eliminate diff and patches functions 
call mdiff variants directly
 2007-07-23 20:44:08 -05:00
Matt Mackall
9257ef4f4f revlog: speed up chunkcache 
- use a reasonable cache size
- avoid an extra copy when we pull in big revs
 2007-07-23 20:44:08 -05:00
Matt Mackall
141c035280 revlog: some basic code reordering 2007-07-23 20:44:08 -05:00
Matt Mackall
ff72ed852c revlog: abstract out index entry packing 
This lets us remove most of the remaining v0/v1 logic from revlog
 2007-07-23 20:44:08 -05:00
Matt Mackall
1f16099cd0 revlog: simplify revlog.__init__ 
- move stat into io helper
- get rid of self.defversion and self.indexstat
- fold _load into __init__
 2007-07-23 20:44:08 -05:00
Matt Mackall
f3686b0a39 revlog: mark cache private 2007-07-23 20:44:08 -05:00
Matt Mackall
9a71ba31c4 revlog: pull chunkcache back into revlog 2007-07-23 20:44:08 -05:00
Matt Mackall
35c14c89a6 revlog: change _inline from a function to a variable 2007-07-23 20:44:08 -05:00
Matt Mackall
b51cdea681 revlog: simplify addrevision 
- remove unused defaults for p1, p2, and text
- reduce some if/else
- use better variable names
- remove some extra variables
- remove some obsolete corner tests
- simply first entry handling for revlogng
- simply inline vs outofline writeout
 2007-07-23 20:44:08 -05:00
Matt Mackall
0ab56eac5f revlog: some codingstyle cleanups 2007-07-23 20:44:08 -05:00
Matt Mackall
2cb4c11520 revlog: add a magic null revision to our index 
We expand our index by one entry so that index[nullrev] points to a
unique entry, the null revision. This naturally eliminates numerous
extra tests in the performance-sensitive index access functions, most
of which are now trivial again.

Adding new entries is now done with insert(-1, e) rather than
append(e).
 2007-07-23 20:44:08 -05:00
Matt Mackall
f03d3e07aa revlog: change accesses to index entry elements to use positive offsets 2007-07-23 20:44:08 -05:00
Matt Mackall
04b4378f47 revlog: parse revlogv0 indexes into v1 internally 
This lets us eliminate lots of special case code in our
performance-sensitive index accessors.
 2007-07-23 20:44:08 -05:00
Matt Mackall
70e7477200 revlog: only allow lazy parsing with revlogng files 
This will allow us to store indices in memory in a single entry format
 2007-07-23 20:44:08 -05:00
Matt Mackall
214f50b511 revlog: simplify the v1 immediate parser 
- read all the data at once (large files are handled by the lazy parser)
- cache the entire file for inline revlogs
- simplify looping
 2007-07-23 20:44:08 -05:00
Matt Mackall
3b9167f022 revlog: set the threshold for lazy parsing higher 
A typical machine can parse a 1MB index in well under a second
 2007-07-23 20:44:08 -05:00
Matt Mackall
1889aee8b0 revlog: simplify the v0 parser 2007-07-23 20:44:07 -05:00
Matt Mackall
e7ef5dc500 revlog: add revlogio interface 
This lets us separate the logic for different on-disk revlog formats from
the main revlog logic.
 2007-07-23 20:44:07 -05:00
Matt Mackall
2158f34c5f revlog: regroup parsing code 2007-07-23 20:44:07 -05:00
Matt Mackall
d5b7eaa69a revlog: privatize some methods 2007-07-21 16:18:42 -05:00
Matt Mackall
0e54e000c1 revlog: delete unused function makenode 2007-07-21 16:18:24 -05:00
Matt Mackall
9eea27ad83 revlog: raise offset/type helpers to global scope 2007-07-21 16:18:21 -05:00
Thomas Arendsen Hein
2501b7aeac Fix revlog.children so the real children of the null revision can be calculated. 2007-07-03 12:45:33 +02:00
Thomas Arendsen Hein
4d29c6dc8e Updated copyright notices and add "and others" to "hg version" 2007-06-19 08:51:34 +02:00
Alexis S. L. Carvalho
ae21f631dc merge with crew-stable 2007-06-04 00:26:53 -03:00
Alexis S. L. Carvalho
f0513bf1bb Also check the index file size when deciding whether to reload a revlog. 
This should fix some spurious failures from test-clone-pull-corruption
(issue552).
 2007-06-04 00:24:48 -03:00
Alexis S. L. Carvalho
047c4b002b Merge with crew-stable. 2007-04-07 04:45:27 -03:00
Alexis S. L. Carvalho
b98b34fc2e revlog.py: fix/tweak read ahead code in lazyparser 2007-04-07 04:27:55 -03:00
Matt Mackall
746c62bff7 Fixups for recent changes in revlog version handling 2007-04-02 14:16:44 -05:00
Thomas Arendsen Hein
c0c7def517 Fixed spacing in assignment of REVLOG_DEFAULT_VERSION 2007-03-24 13:31:43 +01:00
Matt Mackall
04561e556e revlog: simplify revlog version handling 
- pass the default version as an attribute on the opener
- eliminate config option mess
 2007-03-22 19:52:38 -05:00
Matt Mackall
b4f6965b1d revlog: don't pass datafile as an argument 2007-03-22 19:12:03 -05:00
Matt Mackall
29e8301c27 Merge with stable 2007-03-15 18:57:47 -05:00
Matt Mackall
e33842c05f revlog: attempt to gracefully handle some interleaved damage 2007-03-15 18:38:46 -05:00
Brendan Cully
53a62dfb33 lazyindex: handle __delitem__ in loadblock 2007-02-04 16:08:56 -08:00
Brendan Cully
8a6eea2483 Add revlog.LookupError exception, and use it instead of RevlogError. 
Remove repo.LookupError, which devolves to the revlog version.
 2006-12-18 12:22:43 -08:00
Matt Mackall
887599574c Sync with -stable 2006-12-17 21:48:22 -06:00
Benoit Boissinot
fc997dfbf2 fix calculation of new heads added during push with -r 
fix issue450
 2006-12-17 05:00:22 +01:00
Matt Mackall
296d6a7cb8 Simplify i18n imports 2006-12-14 20:25:19 -06:00
Matt Mackall
f17a4e1934 Replace demandload with new demandimport 2006-12-13 13:27:09 -06:00
Thomas Arendsen Hein
aaa1f20f5a Use unsigned version format. 
This way can use one additional bit, and when encountering invalid revlogs
with the first bit set don't produce python warnings or strange error messages.
 2006-12-01 23:27:53 +01:00
Thomas Arendsen Hein
d279e55ca0 Improve error message for unknown revlog flags. 
Old: unknown flags 120000 for ...
New: unknown flags 0x0012 for ...
 2006-12-01 23:16:12 +01:00
Matt Mackall
b4165617fe Make revlog error slightly less scary 2006-12-01 02:24:56 -06:00
Matt Mackall
9c3965d552 add docstring to reachable 2006-11-19 16:32:36 -06:00
Thomas Arendsen Hein
d3c6666f73 Indentation cleanups for 16c127c30658. 2006-11-19 16:09:04 +01:00
Marcos Chaves
794888ecba fix warnings generated by pygettext.py. 2006-11-18 17:20:38 -02:00
Matt Mackall
aa97044b36 revlog: reachable actually takes a node 2006-11-13 13:26:57 -06:00
Benoit Boissinot
46893fed9e revlog: more nullrev fixes 2006-10-29 16:14:55 +01:00
Benoit Boissinot
a0c2a353ec revlog: return 0 as offset for nullid 2006-10-29 16:14:02 +01:00
Benoit Boissinot
a2d320e112 revlog.size() fix handling of rev == -1 2006-10-29 14:41:14 +01:00
Thomas Arendsen Hein
687e4a7111 Define and use nullrev (revision of nullid) instead of -1. 2006-10-29 10:53:51 +01:00
Alexis S. L. Carvalho
7342f50f68 revlog.py: always return tuples from parents and parentrevs 
There's at least one place doing parentrevs(rev) != (rev-1, -1),
which is always true with revlog v0, since parentrevs returns a list
in this case.
 2006-10-24 23:00:43 -03:00
Matt Mackall
49503fb772 Merge with crew 2006-10-24 13:46:04 -05:00
Benoit Boissinot
7c004b60e9 fix errors spotted by pychecker 2006-10-19 16:55:34 +02:00
Benoit Boissinot
1bccbf1e1e revlog.children: use parentrevs instead of parents 2006-10-19 10:23:27 +02:00
Matt Mackall
4a6017f8e8 Only look up tags and branches as a last resort 2006-10-18 11:44:56 -05:00
Matt Mackall
833cdcbbe2 revlog.lookup tweaks 
- fast path for binary node ids
- direct lookup for full hex ids
- silly change to bin_id
 2006-10-17 22:07:54 -05:00
Alexis S. L. Carvalho
54c70682e5 make revlog.addgroup pass its file handles to addrevision 
This should fix issue255.

It looks like the problem there happens when addgroup calls addrevision
to add a full revision, and addrevision decides to split the index file
into a .i/.d pair.  Since addgroup has an open file handle for the
index file, the renaming of the new .i file to its final name fails on
windows.
 2006-10-14 18:47:48 -03:00
Benoit Boissinot
a1bd76f99b nodesbetween: fix a bug with duplicate heads 2006-10-12 11:40:26 +02:00
Brendan Cully
65be154af0 Teach bdiff to support buffer objects 
manifest.add gives revlog.addrevision a buffer object, which may
be cached and used for a second call in the same session (as mq does
when pushing multiple patches). The other option would be to cast the
buffer to str when caching it.
 2006-10-11 12:06:14 -07:00
Benoit Boissinot
00efcc68ba correctly find the type of 'id' in revlog.lookup 2006-09-30 15:15:59 +02:00
Benoit Boissinot
445628dbea optimize revlog.lookup when passed hex(node)[:...] 
Instead of converting each node from the filenode to a hex form,
convert the arg to a bin form.
For a revlog with 26711 entries, doing 100 lookup:
before: ~18s
after : ~13s
 2006-09-24 13:52:27 +02:00
Benoit Boissinot
1dc0baf068 cleanups in revlog.lookup 
- add comments
- do a clean separation of the different cases
- don't use a list of each possible node when
  doing the lookup, just keep the previous entry
 2006-09-24 13:52:25 +02:00
Matt Mackall
7ca6c6f0dd Switch revlog.ancestor to use revisions rather than nodeids 2006-09-21 18:15:44 -05:00
Matt Mackall
79aed53c85 Abstract ancestor algorithm into generic function 
Make depth calculation non-recursive
Add simple shortcut for linear ancestry

Convert context to use ancestor function
 make memoized parents function

Convert revlog to use ancestor function
 2006-09-20 16:50:50 -05:00
Benoit Boissinot
2caa3e642d fix warnings spotted by pychecker 2006-09-18 17:43:31 +02:00
Alexis S. L. Carvalho
6a3814b114 Avoid negative block sizes in lazyparser. 
This fixes a bug introduced by changeset 605357078795 .
 2006-09-13 14:37:51 -03:00
Alexis S. L. Carvalho
38036387d3 don't let lazyparser read more data than it can handle 
This should fix issue368.
 2006-09-10 19:15:08 -03:00
Matt Mackall
e3e04b8f17 Move cmp bits from filelog to revlog 2006-08-15 14:18:13 -05:00
Vadim Gelfer
dc377b58c1 update copyrights. 2006-08-12 12:30:02 -07:00
Matt Mackall
4dc0a93fd6 revlog.lookup optimization 2006-08-03 21:24:19 -05:00
Benoit Boissinot
490a3977ae correct the handling of linkrev with nullid 2006-07-17 01:58:59 +02:00
Benoit Boissinot
c0ceee40fc lookup should allow -1 to represent nullid (if passed an int as arg) 2006-07-17 01:58:50 +02:00
Vadim Gelfer
e0b0ae3431 clean up trailing white space. 2006-07-12 08:28:00 -07:00
Matt Mackall
ca55a0ae02 revlog: make lookup handle binary nodes 2006-06-28 16:52:42 -05:00
Matt Mackall
e00f914b09 revlog: handle integer arguments to lookup 2006-06-28 15:49:33 -05:00
Alexis S. L. Carvalho
30401d484f Change revlog.heads to walk the revision graph using revision numbers 
On the kernel repo:
$ hg heads -q
           before    after
RevlogNG    1.11     0.52
Revlogv0    0.80     0.69

Since the current code for tags has to find all the heads of the repo,
this also helps there:
$ hg tags
           before    after
RevlogNG    2.35     1.76
Revlogv0    2.04     1.90
 2006-06-20 15:02:23 -03:00
Alexis S. L. Carvalho
6b69fa6621 Add revlog.parentrevs function. 
This allows one to walk the revision graph using only revision numbers,
which can be faster than using revision hashes, especially for
RevlogNG, where the parents of a revision are stored as revision
numbers.
 2006-06-20 14:57:30 -03:00
Thomas Arendsen Hein
5800bfc290 Manifest groups may be empty, so don't abort in this case (fixes issue210). 
Only abort on empty changelog and file revlogs.
 2006-05-31 22:25:20 +02:00
Benoit Boissinot
01540f3159 fix a typo in an error message 2006-05-16 11:07:56 +02:00
Alexis S. L. Carvalho
e2d5b135e9 Fix revlog.parseindex 
parseindex could fail if read returns too little data in the right
moment (e.g. when there's still leftover data from the previous
iteration and read returns less than "s" bytes).
 2006-05-14 18:22:12 -03:00
mason@suse.com
f2b0d07cb4 Reduce ram used for very large inlined index files 
During a clone, an inline index is not converted to a split index
file until the very end.  When the conversion happens, the index
can be very large, and the inline index loading functions always load
the entire index file into ram.

This changes the revlog code to read the index in smaller chunks.
 2006-05-10 15:26:56 -07:00
Vadim Gelfer
20d2979fe8 windows: revlog.lazyparser not always safe to use. 
can not use on windows < nt or if win32 api not available.
 2006-05-10 11:10:18 -07:00
mason@suse.com
58d4ef2538 Use revlogng and inlined data files by default 
This changes revlog specify revlogng by default.  Inlined
data files are also used unless a flags option is found in the .hgrc.
Some example hgrc files:

[revlog]
# use the original revlog format
format=0

[revlog]
# use revlogng.  Because no flags are included, inlined data files
# also be selected
format=1

[revlog]
# use revlogng but do not inline the data files with the index
flags=

[revlog]
# the new default
format=1
flags=inline
 2006-05-08 14:26:18 -05:00
Vadim Gelfer
75ddc9a984 merge with crew. 2006-05-02 14:37:55 -07:00
Vadim Gelfer
fad839fb0b fix file handling bugs on windows. 
add util.posixfile class that has posix semantics on windows.
fix util.rename so it works with stupid windows delete semantics.
 2006-05-02 14:30:00 -07:00
Thomas Arendsen Hein
c9a89940a1 Replaced 0 with REVLOGV0 where this meaning is used. 2006-04-27 21:16:46 +02:00
Thomas Arendsen Hein
4b95cebb2d Corrected error message for incompatible revision flags. 2006-04-27 21:03:07 +02:00