Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | Whitespace changes only | Jon Bratseth | 2016-12-20 | 1 | -5/+3 |
| | |||||
* | Ensure that terms are not empty. | Henning Baldersheim | 2016-12-16 | 2 | -1/+2 |
| | |||||
* | Nonfunctional changes only | Jon Bratseth | 2016-12-13 | 4 | -5/+4 |
| | |||||
* | MapTensor -> MappedTensor | Jon Bratseth | 2016-12-12 | 2 | -3/+0 |
| | |||||
* | Use type when we have it | Jon Bratseth | 2016-12-07 | 1 | -1/+3 |
| | |||||
* | Merge branch 'master' into bratseth/indexed-tensor | Jon Bratseth | 2016-12-07 | 1 | -2/+2 |
|\ | |||||
| * | Revert "Always use 16 distribution bits on hosted systems" | Tor Brede Vekterli | 2016-12-06 | 1 | -2/+2 |
| | | |||||
* | | Indexed tensor implementation | Jon Bratseth | 2016-12-05 | 4 | -15/+13 |
|/ | |||||
* | Correct over-eager renaming | Jon Bratseth | 2016-12-04 | 25 | -41/+41 |
| | |||||
* | Towards always typed, fully specified tensors | Jon Bratseth | 2016-12-04 | 1 | -2/+2 |
| | | | | | | | | | - Tensor addresses do not repeat dimensions. - Tensor addresses must provide a value for all dimensions. - Tensor dimensions are not serialized in JSON (but still are in binary). - Tensor types are required everywhere, except a workaround for JSON deserialization. - Tensor operations are about 50% faster. - Tensor join of two tensors in the same space is about 4000% faster. | ||||
* | Rename method | Jon Bratseth | 2016-12-03 | 25 | -41/+41 |
| | |||||
* | Always use 16 distribution bits on hosted systems | Jon Bratseth | 2016-12-02 | 1 | -2/+2 |
| | | | | | This avoids distribution skew with few bits and massive redistribution events when nodes are deactivated after reducing the number of nodes. | ||||
* | Handle FalseItem in NOT and RANK | Jon Bratseth | 2016-11-30 | 2 | -4/+53 |
| | |||||
* | Handle FalseItem in AND and OR | Jon Bratseth | 2016-11-30 | 2 | -4/+48 |
| | |||||
* | Merge pull request #1221 from yahoo/bratseth/query-sanitation | Henning Baldersheim | 2016-11-30 | 8 | -174/+235 |
|\ | | | | | Bratseth/query sanitation | ||||
| * | Collapse RANK and NOT | Jon Bratseth | 2016-11-30 | 4 | -35/+68 |
| | | |||||
| * | Better testing | Jon Bratseth | 2016-11-30 | 1 | -10/+32 |
| | | |||||
| * | Collapse multilevel AND and OR | Jon Bratseth | 2016-11-30 | 2 | -21/+28 |
| | | |||||
| * | Simplify more | Jon Bratseth | 2016-11-29 | 5 | -44/+43 |
| | | |||||
| * | Clean up some Steinarmess | Jon Bratseth | 2016-11-29 | 5 | -131/+131 |
| | | |||||
* | | Fix tests | Jon Bratseth | 2016-11-29 | 2 | -4/+3 |
|/ | |||||
* | Revert "Revert "Deprecate XML result format and 'templates'"" | Jon Bratseth | 2016-11-18 | 6 | -2/+18 |
| | |||||
* | Revert "Deprecate XML result format and 'templates'" | Henning Baldersheim | 2016-11-18 | 6 | -18/+2 |
| | |||||
* | Deprecate XML result format and 'templates' | Jon Bratseth | 2016-11-18 | 6 | -2/+18 |
| | |||||
* | Merge pull request #1072 from yahoo/vegard/remove-trace-when-tracelevel-is-zero | Vegard Sjonfjell | 2016-11-17 | 1 | -12/+15 |
|\ | | | | | Remove trace when tracelevel is zero | ||||
| * | Remove trace when tracelevel is zero | Vegard Sjonfjell | 2016-11-11 | 1 | -12/+15 |
| | | |||||
* | | Protect against interference from ping threads | Jon Bratseth | 2016-11-14 | 2 | -6/+18 |
|/ | |||||
* | Whitespace changes only | Jon Bratseth | 2016-11-11 | 1 | -4/+5 |
| | |||||
* | Verify the regex pattern upon construction, so the query will fail with ↵ | Henning Baldersheim | 2016-11-11 | 2 | -1/+28 |
| | | | | message to user, instead of corong the backends. | ||||
* | Merge pull request #1057 from yahoo/arnej/remove-internal-reference | Jon Bratseth | 2016-11-10 | 1 | -6/+6 |
|\ | | | | | do not refer to Yahoo-internal technology | ||||
| * | do not refer to Yahoo-internal technology | Arne H Juul | 2016-11-08 | 1 | -6/+6 |
| | | | | | | | | * just explain what actually happens without referring to specific usage | ||||
* | | Let 0.2 be the default value for maxFilterCoverage | Henning Baldersheim | 2016-11-09 | 1 | -1/+1 |
| | | |||||
* | | add totalhits_per_query metric | Arne H Juul | 2016-11-09 | 1 | -0/+2 |
|/ | |||||
* | Correct condition for adding trace message | Jon Bratseth | 2016-11-07 | 1 | -1/+2 |
| | |||||
* | convert RateLimitingSearcher to use simplemetrics | Arne H Juul | 2016-11-03 | 3 | -19/+37 |
| | |||||
* | Merge pull request #988 from ↵ | Jon Bratseth | 2016-11-01 | 1 | -1/+1 |
|\ | | | | | | | | | yahoo/balder/use-same-version-4.5-for-runtime-and-build Unify on version 4.5 on both buil and runtime. | ||||
| * | Unify on version 4.5 on both buil and runtime. | Henning Baldersheim | 2016-11-01 | 1 | -1/+1 |
| | | |||||
* | | Reduce loglevel. | Henning Baldersheim | 2016-11-01 | 1 | -18/+20 |
|/ | | | | | Add braces. Add final/private. | ||||
* | Count invalid and missing queries as failed | Jon Bratseth | 2016-10-26 | 1 | -16/+6 |
| | |||||
* | Merge with master | Jon Bratseth | 2016-10-25 | 1 | -21/+20 |
|\ | |||||
| * | Move messages from log to trace (#925) | Jon Bratseth | 2016-10-21 | 1 | -21/+20 |
| | | |||||
* | | Use a dedicated thread for VIP status | Jon Bratseth | 2016-10-25 | 1 | -4/+3 |
|/ | | | | Also, various nonfunctiuonal changes. | ||||
* | Merge pull request #894 from yahoo/balder/always-provide-coverage | Henning Baldersheim | 2016-10-17 | 14 | -219/+51 |
|\ | | | | | Balder/always provide coverage | ||||
| * | Cleanup tests | Henning Baldersheim | 2016-10-17 | 8 | -138/+38 |
| | | |||||
| * | QF_WARMUP is long gone | Henning Baldersheim | 2016-10-14 | 1 | -1/+0 |
| | | |||||
| * | I think aggregation was removed 6 years ago. And that is the newest of ↵ | Henning Baldersheim | 2016-10-14 | 1 | -13/+0 |
| | | | | | | | | features removed in this pull request. | ||||
| * | Unsupported is unsupported. | Henning Baldersheim | 2016-10-14 | 1 | -4/+0 |
| | | |||||
| * | It has been ignored the last year or so anyway, so removing it. | Henning Baldersheim | 2016-10-14 | 1 | -2/+1 |
| | | |||||
| * | Remove traces of PCODE_MLD_MONITORRESULT and PCODE_MONITORRESULT | Henning Baldersheim | 2016-10-14 | 3 | -42/+4 |
| | | |||||
| * | No value in having an option that is always used. | Henning Baldersheim | 2016-10-14 | 2 | -17/+9 |
| | |