summaryrefslogtreecommitdiffstats
path: root/searchlib/src/tests/attribute
Commit message (Collapse)AuthorAgeFilesLines
* Change tests and reintroduce assert.Henning Baldersheim2018-05-241-1/+1
|
* Renaming and constification after code review comments.Henning Baldersheim2018-05-241-2/+2
|
* Only return the weight of the first match. Accumulate on the outside.Henning Baldersheim2018-05-231-32/+41
|
* Ensure conformance tests pass.Henning Baldersheim2018-05-231-12/+25
|
* Refactored to allow iteration over matching elements.Henning Baldersheim2018-05-231-7/+7
|
* Remove redundant typedefHenning Baldersheim2018-05-111-1/+0
|
* Adjust naming: referenced lid -> target lid.Tor Egge2018-04-261-44/+44
|
* Test that isImported() returns true.Geir Storli2018-04-231-0/+4
|
* Remove unused AttributeEnumGuard.Tor Egge2018-04-201-1/+1
|
* Change AttributeContext to use makeReadGuard method on attribute vectors.Tor Egge2018-04-203-13/+22
|
* Change target attribute type in ImportedAttributeVector toTor Egge2018-04-201-1/+1
| | | | ReadableAttributeVector.
* Add type-safe down-cast to ITensorAttribute in IAttributeVector and use this ↵Geir Storli2018-04-191-2/+7
| | | | instead of dynamic_cast.
* Add ReadableAttributeVector interface used to create a short-lived read ↵Geir Storli2018-04-192-1/+2
| | | | | | guard over an attribute vector. Let AttributeVector and ImportedAttributeVector implement this interface.
* Remove useReadGuard template argument from fixtures as read guard is always ↵Geir Storli2018-04-191-104/+18
| | | | used now.
* Change ImportedAttributeVector to no longer implement IAttributeVector API.Geir Storli2018-04-192-31/+26
| | | | All read access happens through short lived ImportedAttributeVectorReadGuard.
* make the test pass again as the code was already correct.Henning Baldersheim2018-04-031-1/+1
|
* Add test to prove that the filter aspect is correctly handled also for the ↵Henning Baldersheim2018-04-031-1/+22
| | | | DirectAttributeBlueprint
* Cleanup formatting.Henning Baldersheim2018-03-241-32/+18
|
* Rewrite test to new test syntaxHenning Baldersheim2018-03-241-61/+25
|
* Improve buffer allocation strategy in data store by matching underlying ↵Geir Storli2018-01-261-11/+20
| | | | | | | | allocators. This should reduce the amount of memory wasted in allocations. 1) heap allocation: buffer size is power of 2 to match vespamalloc. 2) mmap allocation: buffer size is multiple of huge page size (2MB) to match mmap allocator.
* Keep imported tensor attribute vector read guard live during access.Tor Egge2018-01-231-1/+2
|
* Revert "Revert "Add imported tensor attribute vector""Tor Egge2018-01-231-0/+97
|
* Revert "Add imported tensor attribute vector"Håkon Hallingstad2018-01-231-97/+0
|
* Merge pull request #4733 from ↵Tor Brede Vekterli2018-01-231-0/+97
|\ | | | | | | | | vespa-engine/toregge/add-imported-tensor-attribute-vector Add imported tensor attribute vector
| * Use ASSERT_TRUE instead of EXPECT_TRUE.Tor Egge2018-01-231-1/+1
| | | | | | | | Continuing on failure would lead to test program crashing.
| * Fix typo in tensor spec.Tor Egge2018-01-221-1/+1
| |
| * Add imported tensor attribute vector read guard.Tor Egge2018-01-221-0/+97
| | | | | | | | | | Extend unit tests for imported attribute vector to check imported tensor attribute.
* | Add config for the alloc grow factor used when allocating buffers in ↵Geir Storli2018-01-222-7/+7
| | | | | | | | multi-value attributes.
* | Add more conservative buffer allocation strategy in datastore buffers.Geir Storli2018-01-191-2/+8
|/ | | | | | | | | | | This should greatly reduce the memory footprint of multi-value attribute vectors which uses an array store for underlying values. The size of a new buffer is calculated as (simplified): *) before: (size of previous buffer) * 2 *) now: (size of all previous buffers) * allocation grow factor (default = 0.2) In addition we now support resizing of all buffers, not only the first one.
* Eliminate mutated version of Runnable.Tor Egge2017-12-124-61/+17
|
* Use standard locking in searchlib (pass 1).Tor Egge2017-12-122-13/+13
|
* update external code (getType->type)Håvard Pettersen2017-11-281-3/+3
|
* Make attribute disk usage estimate more robust. Use cached values insteadTor Egge2017-09-211-0/+1
| | | | of calling methods that require an attribute guard being held.
* Keep a document meta store read guard together with bitvector posting list ↵Geir Storli2017-09-112-2/+5
| | | | | | in search cache for imported attributes. This is to ensure that no lids that are cached in the bitvector are re-used until the guard is released.
* Use bit vector search cache in imported search context if available.Geir Storli2017-09-062-21/+90
| | | | | Try to use bit vector from cache for the given query term if found. Insert bit vector posting list into cache after calculation.
* Implement simple bitvector search cache for use in imported attribute vector.Geir Storli2017-09-062-0/+69
|
* Test that reference attribute tracks unique gids.Tor Egge2017-09-031-0/+34
|
* Keep track of mapping from gid to referenced lid in reference attributeTor Egge2017-08-311-35/+6
| | | | | | even when no documents currently reference that gid in the reference attribute. This eliminates the need for using the gid to lid mapper after populateReferencedLids() has been called.
* Propagate API changes to GidToLidChangeListener and ReferenceAttribute:Tor Egge2017-08-301-12/+15
| | | | Split put and remove notifications into separate methods.
* If fetchPostings() is called with strict set to true then buildTor Egge2017-08-241-1/+8
| | | | | | | | synthetic posting list based on reverse mapping and result of scanning target attribute (via target search context and corresponding search iterator). Use synthetic posting list if available when creating iterator.
* Explicit `this` not needed for own type memberTor Brede Vekterli2017-08-241-1/+1
|
* Map child LID to parent LID space in onSerializeFor...Sort functionsTor Brede Vekterli2017-08-241-16/+37
|
* Adjust test names.Tor Egge2017-08-231-2/+2
|
* Trim down unit test.Tor Egge2017-08-231-116/+5
| | | | | | | | Test adding info from multiple posting lists to bit vector. Make internal merge method private. Move comment.
* Add unit test for posting list merger.Tor Egge2017-08-232-0/+275
|
* Add string from enum lookup to IAttributeVectorTor Brede Vekterli2017-08-211-0/+22
|
* Merge pull request #3158 from ↵Geir Storli2017-08-211-23/+101
|\ | | | | | | | | vespa-engine/toregge/add-imported-attribute-vector-read-guard Add ImportedAttributeVectorReadGuard, used to handle imported attributes
| * Add ImportedAttributeVectorReadGuard, used to handle imported attributesTor Egge2017-08-181-23/+101
| | | | | | | | via ImportedAttributesContext in a safe manner.
* | Remove more clutter in fastos/types.hHenning Baldersheim2017-08-181-2/+1
|/
* Factor out ReferenceMappings from ReferenceAttribute.Tor Egge2017-08-151-2/+3
|