summaryrefslogtreecommitdiffstats
path: root/searchcore/src/tests
Commit message (Collapse)AuthorAgeFilesLines
* Migrate document counts metrics to new naming scheme.Geir Storli2018-09-211-2/+2
| | | | Also make terminology more clear with 'active', 'ready', 'total', 'removed'.
* Revert "Migrate document counts metrics to new naming scheme."Henning Baldersheim2018-09-201-2/+2
|
* Migrate document counts metrics to new naming scheme.Geir Storli2018-09-201-2/+2
| | | | Also make terminology more clear with 'active', 'ready', 'total', 'removed'.
* Must wait longer than stats update interval.Henning Baldersheim2018-09-191-1/+1
|
* Remove remains of spooler code from searchcore.Tor Egge2018-09-141-12/+0
|
* Add using HeapSize to avoid long namespace qualifiersLester Solbakken2018-09-061-1/+2
|
* Don't repeat rerank-count overrideLester Solbakken2018-09-051-1/+1
|
* Revert "Bratseth/handle large constants take 3"Jon Bratseth2018-09-052-2/+2
|
* Move ranking-constants.def to configdefinitionsJon Bratseth2018-09-052-2/+2
|
* Ignore GC flush targets when calculating oldest flushed serial number.Geir Storli2018-09-041-16/+50
| | | | | The oldest flushed serial number is used to prune the transaction log up to this point in time. As GC flush targets are not dependent on data stored in the transaction log they should not affect pruning.
* Simplify and modernize test code.Geir Storli2018-09-041-92/+77
|
* Rename test file for flush engine.Geir Storli2018-09-042-1/+1
|
* Revert "Geirst/fix calc of oldest flushed serial in flush engine"Jon Bratseth2018-09-042-127/+108
|
* Ignore GC flush targets when calculating oldest flushed serial number.Geir Storli2018-09-031-16/+50
| | | | | The oldest flushed serial number is used to prune the transaction log up to this point in time. As GC flush targets are not dependent on data stored in the transaction log they should not affect pruning.
* Simplify and modernize test code.Geir Storli2018-09-031-92/+77
|
* Rename test file for flush engine.Geir Storli2018-09-032-1/+1
|
* Expose constness in interface to avoid cheating.Henning Baldersheim2018-08-291-1/+1
|
* Wire in the attribute execution via the IAttributeContext.Henning Baldersheim2018-08-2710-114/+24
| | | | Also execution in a task for containment.
* - Add option to increment an attribute for every doc matched.Henning Baldersheim2018-08-271-5/+13
| | | | | | | | | | | - Add option to increment an attribute for every doc reranked. - Add option to increment an attribute for every doc fetched. Conflicts: searchcore/src/vespa/searchcore/proton/matching/attribute_operation.cpp Conflicts: searchlib/src/vespa/searchlib/attribute/i_attribute_functor.h
* Merge pull request #6636 from vespa-engine/balder/add-attribute_operationHenning Baldersheim2018-08-231-2/+2
|\ | | | | Balder/add attribute operation
| * Move AttributeOperation and AttributeFunctor to searc::attributeHenning Baldersheim2018-08-223-151/+2
| |
| * Test non-updateable attributes.Henning Baldersheim2018-08-221-9/+36
| |
| * Test all operations over both integer and floating point attributes.Henning Baldersheim2018-08-201-2/+77
| |
| * Add tests for operation parsingHenning Baldersheim2018-08-202-0/+47
| |
* | Use std::make_xxxHenning Baldersheim2018-08-203-22/+14
|/ | | | | nested namespace And various other style changes.
* Remove attribute vector saveAs method with two arguments.Tor Egge2018-08-171-3/+3
|
* adjust rank score re-scaling when diversifyingHåvard Pettersen2018-08-161-17/+40
| | | | | implicitly boost the pre re-rank score of selected results based on the score of the best result dropped by the diversity filter.
* use SortedHitSequence when selecting best results among threadsHåvard Pettersen2018-08-141-15/+24
|
* use non-const copy of input in rendezvousHåvard Pettersen2018-08-131-3/+3
|
* Pass stringref by valueHenning Baldersheim2018-08-111-2/+2
|
* Ensure object lifetime is upheld.Henning Baldersheim2018-08-101-60/+41
|
* Separate test to be more clear what is being tested.Henning Baldersheim2018-08-101-4/+13
|
* get_mtf -> create_mtfHenning Baldersheim2018-08-101-11/+11
|
* Add a test for diversity after first phase.Henning Baldersheim2018-08-091-3/+91
|
* Create diversity filter on demand if diversity is specified.Henning Baldersheim2018-08-091-5/+15
|
* Merge pull request #6485 from ↵Henning Baldersheim2018-08-081-23/+46
|\ | | | | | | | | vespa-engine/balder/transfer-when-selecting-the-best Balder/transfer when selecting the best
| * Improve tests by tightening it.Henning Baldersheim2018-08-081-17/+16
| |
| * Add an optional diversification filter that is applied inline when selecting ↵Henning Baldersheim2018-07-271-0/+14
| | | | | | | | the best.
| * Select the hits to rerank in the selectBest part instead of a later copy.Henning Baldersheim2018-07-271-23/+33
| | | | | | | | Stick to full hits instead of feature_t.
* | Sync attribute dirs before marking attribute snapshot valid.Tor Egge2018-08-032-1/+5
| |
* | Remove unused document db dirs.Tor Egge2018-07-312-0/+187
| |
* | Rename method according to review feedback.Tor Egge2018-07-271-1/+1
| | | | | | | | Remove unneeded guard checks.
* | Add skeleton for proton disk layout, to better handle removal ofTor Egge2018-07-271-14/+61
|/ | | | unused directories in proton directory tree.
* Merge pull request #6469 from ↵Tor Brede Vekterli2018-07-262-2/+2
|\ | | | | | | | | vespa-engine/vekterli/remove-dangerous-stringref-c_str-function Remove dangerous stringref::c_str()
| * Remove stringref::c_str()Tor Brede Vekterli2018-07-252-2/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | The expected semantics of c_str() (a null-terminated string) cannot be satisfied with a string reference, so remove the function entirely to prevent people from using it in buggy ways. Replaces c_str() with data() in places where it is presumed safe, otherwise constructs temporary string instances. Certain callsites have been de-stringref'd in favor of regular strings, in particular where C APIs have been transitively called. The vast majority of these were called with string parameters anyway, so should not cause much extra allocation.
* | Track document db destruction.Tor Egge2018-07-251-10/+10
|/ | | | Wait until document db has been destroyed when pruning old document dbs.
* Move the whitelist down through AndNot and Rank to the last one and 'AND' it ↵Henning Baldersheim2018-07-241-5/+12
| | | | with the first child there.
* Let AndNot stay on top after whitelisting too.Henning Baldersheim2018-07-231-4/+17
|
* Add a test that RankItems stay on top after whitelisting.Henning Baldersheim2018-07-201-0/+38
|
* Actually test what you claimHenning Baldersheim2018-07-191-2/+2
|