Repository navigation
Use union with bind joins as physical (sub)plan for a join with an mreq operator - #759
Merged
Merged
Conversation
hartig
added a commit
that referenced
this pull request
Oct 10, 2026
Merged
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
So far, the logical-to-physical plan converter converted a join between two multi-request (mreq) operators into a symmetric hash join (SHJ) over the two (physical) mreq operators. This PR changes this behavior: Now the resulting physical plan is a union over bind-joins created from the second mreq operator; the first mreq operator is used as input to to all these bind joins (i.e., the physical plan is not a tree anymore but a DAG).
@theray1 This is not yet the new multi-source bind-join algorithm that I am planning to implement, but just a rewriting using the traditional single-source bind-join algorithm, exploiting the fact that HeFQUIN can pipe the output of one operator (the first mreq operator, in this case) as input to multiple operators (in this case, all the bind joins for all the federation members of the second mreq operator). When testing with the
q05a_jou_public.sparqlquery that you gave me, I see a reduction of the plan execution time from around 30,000 ms (=30 secs) for the previous SHJ-based plan down to about 1,000 ms for the bind-join-based plan!!! :-)/cc @keski