Improved ql:has-relation versatility. #72
Conversation
floriankramer
force-pushed
the
hasrelation
branch
from
0ea60c5
to
a82fbbe
Compare
|
Can you add queries covering the different uses of |
src/engine/HasRelationScan.cpp
Outdated
| os << " "; | ||
| } | ||
| switch (_type) { | ||
| // TODO(florian): these are not good cache keys |
There was a problem hiding this comment.
You mean the tree->asString() in general or these specifically?
There was a problem hiding this comment.
HasRelationScan's keys originally contained variable names, I'd fixed that but apparently forgot to remove the TODO.
src/engine/HasRelationScan.cpp
Outdated
| } | ||
|
|
||
| size_t HasRelationScan::getSizeEstimate() { | ||
| // TODO: these size estimates only work if all predicates are functional |
There was a problem hiding this comment.
which they aren't right? So is this a problem?
There was a problem hiding this comment.
I apparently forgot to fix this after adding the computation of the has-relation size and multiplicities to the index. Replaced the rather wild guesses used before with the apporximations used by the join operation.
src/engine/HasRelationScan.cpp
Outdated
| new std::vector<std::array<Id, 1>>(); | ||
| result->_fixedSizeData = fixedSizeData; | ||
|
|
||
| size_t id = 0; |
src/engine/HasRelationScan.cpp
Outdated
| } | ||
|
|
||
| template <typename T, typename C> | ||
| struct resizeIfVec { |
There was a problem hiding this comment.
This needs comments explaining why and where this is used
| static_cast<vector<array<Id, InputColCount + 1>>*>( | ||
| result->_fixedSizeData), | ||
| hasPattern, hasRelation, patterns); | ||
| } else { |
There was a problem hiding this comment.
Add a comment to make it clear that this is only for the == 5 case where the input uses _fixedSizeData still but the output already uses _varSizeData.
src/engine/QueryPlanner.cpp
Outdated
|
|
||
| std::shared_ptr<QueryExecutionTree> hasRelationScan( | ||
| new QueryExecutionTree(_qec)); | ||
| std::shared_ptr<QueryExecutionTree> other( |
| @@ -424,7 +424,7 @@ string Server::composeResponseJson(const string& query, | |||
|
|
|||
| string msg = ad_utility::escapeForJson(exception.what()); | |||
|
|
|||
| os << "\"Exception-Error-Message\": \"" << msg << "\"\n" | |||
| os << "\"exception\": \"" << msg << "\"\n" | |||
There was a problem hiding this comment.
Do the QLever UI folks know about this change already?
There was a problem hiding this comment.
I hadn't realized this could cause problems for them. I changed the name to be consistent with the ad_semsearch::Exception naming, as the old ui does not print the error message otherwise.
src/index/Index.cpp
Outdated
| file.write(entityHasPattern.data(), sizeof(Id) * numHasPatterns * 2); | ||
|
|
||
| // write the entityHasRelation vector | ||
| size_t numHasRelation = entityHasRelation.size(); |
There was a problem hiding this comment.
numHasRelations (add s) to make it consistent with numHasPatterns
| file.close(); | ||
|
|
||
| hasPattern.resize(entityHasPattern.back()[0] + 1); | ||
| // hasRelation.resize(entityHasRelation.back()[0] + 1); | ||
| LOG(INFO) << "Done creating patterns file." << std::endl; |
There was a problem hiding this comment.
Maybe also print how many patterns we found
There was a problem hiding this comment.
I added the number of patterns to the rest of the info a couple of lines above this one.
src/index/Index.cpp
Outdated
| while (buffer[i][0] > pos) { | ||
| _hasPattern[pos] = NO_PATTERN; | ||
| pos++; | ||
| // Store all data in the file |
e2e/scientists_queries.yaml
Outdated
| ?entity ql:has-relation ?r. | ||
| } | ||
| checks: | ||
| # greater than the current limit |
src/engine/QueryPlanner.cpp
Outdated
| // Check if one of the two operations is a HAS_RELATION_SCAN . | ||
| // If the join column correspods the the has-relation scans | ||
| // Check if one of the two operations is a HAS_RELATION_SCAN. | ||
| // If the join column corresponds to the the has-relation scan's |
floriankramer
force-pushed
the
hasrelation
branch
from
cf9cadc
to
6543351
Compare
ql:has-relation can now be used like any other predicate, with the pattern trick still being used to answer queries matching its specific requirements.
I also ran clang-format on the entire repository which changed a handful of files and added those in the last commit of this pr.