Commit graph

309 commits

Author SHA1 Message Date
e8c51f5df4 Merge kortschak/config (hand-conflict) 2014-08-23 17:21:08 -04:00
c46c3f16e0 Merge pull request #138 from kortschak/usage
Clean up usage
2014-08-23 14:20:33 -04:00
79297cd8b1 Merge pull request #141 from kortschak/parse
Fix parser acceptance of `"` literal in IRIRef
2014-08-23 14:18:43 -04:00
kortschak
49961a50d7 Generate parser and add tests
ragel -Z -G2 parse.rl

ragel version 6.8
2014-08-23 11:09:28 +09:30
kortschak
d7e4aff72d Fix IRIRef defintions
Literal `"` is not allowed in an IRIRef. Fix this.
2014-08-23 10:57:49 +09:30
kortschak
d9096d6d9f Make usage cayley intro banner contextual
Only provide it when people run cayley without any other args.
2014-08-22 16:40:22 +09:30
kortschak
318edfc3c7 Clean up usage
* Use raw strings.
* Hook usage into flag.
* Print banner to stderr as flag.PrintDefaults does.
* Use the cayley usage rather than the flag ussage when command is
  unknown.
* Simplify args handling.
2014-08-22 16:33:20 +09:30
kortschak
df2c5e3c2a Move flag handling out of config into main
This places all the flag definitions together, making them easier to
find, and makes it possible (in the future) to use db for convenience
functions when we have a Go API, without having flag space contaminated
by cayley main flags.
2014-08-22 09:56:36 +09:30
cce0f88803 Merge pull request #135 from kortschak/mongo
Quieten go vet in mongo
2014-08-21 13:27:27 -04:00
c8c7f1ff8c Merge pull request #132 from kortschak/repl
Fix error in REPL term shutdown leaving tty unsane
2014-08-21 13:26:54 -04:00
ce25bd9b76 Merge pull request #131 from kortschak/tagger
Do tagger copying with less iteration
2014-08-21 13:23:41 -04:00
kortschak
93c9814147 Quieten go vet in mongo 2014-08-21 19:33:00 +09:30
kortschak
631188c626 Fix error in REPL term shutdown leaving tty unsane
Previously we did not close the liner term unless a SIGINT or SIGKILL
was received. This left the terminal in raw. Fix that.
2014-08-21 15:55:05 +09:30
kortschak
08e47b4a9e Do tagger copying with less iteration
This makes minimal difference to the benchmarks in cayley_test.go (a
variable ±5% on the bigger cases).
2014-08-21 15:34:50 +09:30
775a253a06 Merge pull request #127 from barakmich/localhost
Switch default to localhost. Fixes #107
2014-08-20 16:20:05 -04:00
b4ef86145d Switch default to localhost. Fixes #107 2014-08-20 15:56:36 -04:00
720a80a51f Merge pull request #126 from barakmich/replcrash
Temporarily fix REPL crash by locking the kill channel
2014-08-20 14:01:25 -04:00
d3d55e320d Merge pull request #125 from barakmich/gremlin_underscore
Import underscore for Gremlin
2014-08-20 13:52:05 -04:00
8e16543da3 Merge branch 'master' into replcrash 2014-08-20 13:44:24 -04:00
f5f0ed7173 Import gremlin, v=2 for more cases 2014-08-20 13:39:26 -04:00
2eb25968c2 Merge pull request #122 from barakmich/explain
Explain Level Logging and Correcting the Cost Model
2014-08-20 01:06:21 -04:00
ca6922e79b fix the tests and care not about result ordering 2014-08-20 01:02:52 -04:00
c1ff6ce1aa lock s.kill, send done signal 2014-08-16 18:03:05 -04:00
f605e1138d kortschak's fix 2014-08-16 16:52:09 -04:00
a371155cd2 Fix linksto and zero-costs 2014-08-16 06:30:27 -04:00
e453385d5e explain logging and weight fixing 2014-08-16 05:19:16 -04:00
e1e95b9686 Merge pull request #113 from barakmich/log_database
Convert Cayley indexing to an append-only log
2014-08-16 03:26:49 -04:00
0ffb244b7b add test dep for travis 2014-08-14 21:40:19 -04:00
3b83845584 Merge pull request #1 from kortschak/b
Use cznic/b B+tree implementation in place of GoLLRB for memstore
2014-08-14 21:39:42 -04:00
8720e17d87 Merge hash pool in from master
Conflicts:
	graph/leveldb/triplestore.go
	graph/mongo/triplestore.go
2014-08-14 21:03:36 -04:00
55b03ce4c1 Merge pull request #120 from barakmich/pool
Use sync.Pool to hold hashing objects for KV backends
2014-08-14 20:58:25 -04:00
de882b9f93 port hasher pool to leveldb 2014-08-14 20:47:55 -04:00
kortschak
c618e556f4 Merge branch 'log_database' into b
Comparison of b against GoLLRB (as at d5f020).

$ benchcmp gollrb.bench b-gen.bench
benchmark                                   old ns/op       new ns/op	delta
BenchmarkNamePredicate                      1631932         1409531	-13.63%
BenchmarkLargeSetsNoIntersection            190792654       63748682	-66.59%
BenchmarkVeryLargeSetsSmallIntersection     896154437       373475843	-58.32%
BenchmarkHelplessContainsChecker            20719182678     14078301640	-32.05%
BenchmarkNetAndSpeed                        32519019        20188665	-37.92%
BenchmarkKeanuAndNet                        18319247        15224988	-16.89%
BenchmarkKeanuAndSpeed                      30849568        18744134	-39.24%
BenchmarkKeanuOther                         105552525       107620648	+1.96%
BenchmarkKeanuBullockOther                  295395338       115193002	-61.00%

benchmark                                   old allocs     new allocs	delta
BenchmarkNamePredicate                      1339           1341	+0.15%
BenchmarkLargeSetsNoIntersection            22585          23632	+4.64%
BenchmarkVeryLargeSetsSmallIntersection     65776          69396	+5.50%
BenchmarkHelplessContainsChecker            1713541        2036316	+18.84%
BenchmarkNetAndSpeed                        17104          17240	+0.80%
BenchmarkKeanuAndNet                        15816          15855	+0.25%
BenchmarkKeanuAndSpeed                      16368          16493	+0.76%
BenchmarkKeanuOther                         30134          30634	+1.66%
BenchmarkKeanuBullockOther                  35510          36454	+2.66%

benchmark                                   old bytes     new bytes	delta
BenchmarkNamePredicate                      96162         96294	+0.14%
BenchmarkLargeSetsNoIntersection            1172356       1249872	+6.61%
BenchmarkVeryLargeSetsSmallIntersection     2810080       2992409	+6.49%
BenchmarkHelplessContainsChecker            89233264      104999088	+17.67%
BenchmarkNetAndSpeed                        1388793       1428110	+2.83%
BenchmarkKeanuAndNet                        1263145       1250079	-1.03%
BenchmarkKeanuAndSpeed                      1246956       1281546	+2.77%
BenchmarkKeanuOther                         2021312       2024727	+0.17%
BenchmarkKeanuBullockOther                  2671448       2742968	+2.68%

Conflicts:
	graph/memstore/triplestore.go
2014-08-15 10:17:52 +09:30
kortschak
737037a894 Experiment with sync.Pool 2014-08-14 19:03:55 +09:30
d5f020bc2e Merge branch 'master' into log_database 2014-08-14 02:36:10 -04:00
6d609f191a Merge pull request #119 from barakmich/go13
Bump to Go 1.3
2014-08-14 02:35:24 -04:00
02f12834d7 up to go 1.3 2014-08-14 02:31:00 -04:00
d2026ea271 fix mongo indexing name mismatch 2014-08-14 01:57:03 -04:00
f967b36f84 comments and concretized deltas 2014-08-14 01:24:39 -04:00
fe0569c9d4 merge with master 2014-08-13 23:03:07 -04:00
11cfc3dfd8 Merge pull request #118 from barakmich/issue117
Fixes #117, HasA speedup
2014-08-13 21:23:01 -04:00
cfca7db47c Go style
My C++ programmer's disease was acting up.
2014-08-13 14:54:36 -04:00
03798bc4fa fix mongo hasher 2014-08-13 13:29:08 -04:00
104e7d110d fix leveldb (and speed up hasa) 2014-08-13 13:08:50 -04:00
kortschak
4a92ae9758 Merge branch 'log_database' into b
Conflicts:
	graph/memstore/iterator.go
	graph/memstore/triplestore.go
2014-08-13 17:15:11 +09:30
kortschak
d98ca99974 Merge branch 'master' into b
$ benchcmp gollrb.bench b-gen.bench
benchmark                                   old ns/op       new ns/op	delta
BenchmarkNamePredicate                      1369329         1444990	+5.53%
BenchmarkLargeSetsNoIntersection            72329029        64975716	-10.17%
BenchmarkVeryLargeSetsSmallIntersection     890824761       408784476	-54.11%
BenchmarkHelplessContainsChecker            35314797618     30673240485	-13.14%
BenchmarkNetAndSpeed                        19694146        19486797	-1.05%
BenchmarkKeanuAndNet                        15340756        15317415	-0.15%
BenchmarkKeanuAndSpeed                      17902709        18042030	+0.78%
BenchmarkKeanuOther                         53452058        50984817	-4.62%
BenchmarkKeanuBullockOther                  90827780        86536510	-4.72%

benchmark                                   old allocs     new allocs	delta
BenchmarkNamePredicate                      1339           1339		+0.00%
BenchmarkLargeSetsNoIntersection            22603          22674	+0.31%
BenchmarkVeryLargeSetsSmallIntersection     65787          65860	+0.11%
BenchmarkHelplessContainsChecker            1713541        1713669	+0.01%
BenchmarkNetAndSpeed                        17135          17146	+0.06%
BenchmarkKeanuAndNet                        15802          15802	+0.00%
BenchmarkKeanuAndSpeed                      16397          16396	-0.01%
BenchmarkKeanuOther                         30148          30149	+0.00%
BenchmarkKeanuBullockOther                  35542          35544	+0.01%

benchmark                                   old bytes     new bytes	delta
BenchmarkNamePredicate                      96226         95842		-0.40%
BenchmarkLargeSetsNoIntersection            1165914       119725	+2.69%
BenchmarkVeryLargeSetsSmallIntersection     2760072       2777798	+0.64%
BenchmarkHelplessContainsChecker            84388448      84351168	-0.04%
BenchmarkNetAndSpeed                        1414837       1425752	+0.77%
BenchmarkKeanuAndNet                        1247249       1247453	+0.02%
BenchmarkKeanuAndSpeed                      1275522       1275243	-0.02%
BenchmarkKeanuOther                         2021107       2021497	+0.02%
BenchmarkKeanuBullockOther                  2682243       2683250	+0.04%
2014-08-13 16:01:32 +09:30
1b22203e7a Merge pull request #109 from kortschak/empty
Allow memstore instances to be started empty
2014-08-12 19:42:46 -04:00
ee11a26bc2 Merge pull request #116 from barakmich/issue114
Fix Issue 114 -- And subiterators may move forward on Contains, even if the And ultimately doesn't
2014-08-12 19:32:30 -04:00
50efe31fdd fix the test, remove -short from .travis.yml 2014-08-12 19:26:02 -04:00
b6d966e9b5 fix and iterator, add some logging, and reenable 2014-08-12 16:38:10 -04:00