-
Notifications
You must be signed in to change notification settings - Fork 3.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
sql: move indexJoiner into joinReader #40749
Comments
5 tasks
asubiotto
changed the title
sql: improve join reader interface
sql: move indexJoiner into joinReader
Jun 2, 2020
We're planning on getting rid of the |
Yeah, that'd be great! |
We would probably want to preserve the EXPLAIN output for index join, so we may want to plumb a flag just for cosmetic purposes. |
helenmhe-zz
pushed a commit
to helenmhe-zz/cockroach
that referenced
this issue
Jul 15, 2020
Previously there was a separate processor specifically for index joins. This commit instantiates a joinReader processor instead, so that we can get rid of the indexJoiner processor and indexJoinerNode. Resolves: cockroachdb#40749 Release note: None
helenmhe-zz
pushed a commit
to helenmhe-zz/cockroach
that referenced
this issue
Jul 15, 2020
Previously there was a separate processor specifically for index joins. This commit instantiates a joinReader processor instead, so that we can get rid of the indexJoiner processor and indexJoinerNode. Resolves: cockroachdb#40749 Release note: None
helenmhe-zz
pushed a commit
to helenmhe-zz/cockroach
that referenced
this issue
Jul 16, 2020
Previously there was a separate processor specifically for index joins. This commit instantiates a joinReader processor instead, so that we can get rid of the indexJoiner processor and indexJoinerNode. Resolves: cockroachdb#40749 Release note: None
helenmhe-zz
pushed a commit
to helenmhe-zz/cockroach
that referenced
this issue
Jul 16, 2020
Previously there was a separate processor specifically for index joins. This commit instantiates a joinReader processor instead, so that we can get rid of the indexJoiner processor and indexJoinerNode. Resolves: cockroachdb#40749 Release note: None
helenmhe-zz
pushed a commit
to helenmhe-zz/cockroach
that referenced
this issue
Jul 21, 2020
Previously there was a separate processor specifically for index joins. This commit instantiates a joinReader processor instead, so that we can get rid of the indexJoiner processor and indexJoinerNode. Resolves: cockroachdb#40749 Release note: None
helenmhe-zz
pushed a commit
to helenmhe-zz/cockroach
that referenced
this issue
Jul 22, 2020
Previously there was a separate processor specifically for index joins. This commit instantiates a joinReader processor instead, so that we can get rid of the indexJoiner processor and indexJoinerNode. Resolves: cockroachdb#40749 Release note: None
helenmhe-zz
pushed a commit
to helenmhe-zz/cockroach
that referenced
this issue
Jul 22, 2020
Previously there was a separate processor specifically for index joins. This commit instantiates a joinReader processor instead, so that we can get rid of the indexJoiner processor and indexJoinerNode. Resolves: cockroachdb#40749 Release note: None
craig bot
pushed a commit
that referenced
this issue
Jul 28, 2020
51183: geogen: introduce package to generate random geo types r=sumeerbhola a=otan Generate random geo types in a new geogen package. This is immediately utilised for all the randomized testing, including sqlsmith. Release note: None 51240: sql: implement the `DROP TYPE` command r=rohany a=rohany This PR implements the `DROP TYPE` command without support for `CASCADE`. It lays the foundation of the dependency management between types and tables that use them. With this, we can drop types when a database is dropped. Fixes #48363. Release note: None 51416: rowexec: refactor index join to use JoinReader r=helenmhe a=helenmhe Previously there was a separate processor specifically for index joins. This commit instantiates a joinReader processor instead, so that we can get rid of the indexJoiner processor and indexJoinerNode. Resolves: #40749 Release note: None 51537: builtins: implement ST_Intersection and ST_Buffer for geography types r=sumeerbhola a=otan PostGIS uses _ST_BestSRID to approximate a geography as a geometry to perform ST_Buffer and ST_Intersection. We're obliged to do the same - comments are inline. Resolves #48812 Resolves #48389 Resolves #48390 Resolves #48391 Resolves #48398 Release note (sql change): Implements the ST_Buffer and ST_Intersection functions for Geography types. Release note (sql change): Implement ST_Intersection for string types. 51923: opt: move code that generates proto explain tree r=RaduBerinde a=RaduBerinde Move the code that generates the `roachpb.ExplainTreePlanNode` tree (shown in the UI) to exec/explain. It parallels the existing code for generating other explain output. Release note: None Co-authored-by: Oliver Tan <otan@cockroachlabs.com> Co-authored-by: Rohan Yadav <rohany@alumni.cmu.edu> Co-authored-by: Helen He <helenhe.mit@gmail.com> Co-authored-by: Radu Berinde <radu@cockroachlabs.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
The join reader has two somewhat distinct "modes" of operation (index join and general lookup join). The index join mode has some limitations: it expects that the input stream starts with the PK columns in order, and is not capable of "passing through" other values - we may have values for other columns available but the index join has to get and decode those again.
As we remove the heuristic planner, we can lift some of the restrictions around
indexJoinNode
(e.g. allow arbitrary (non-scanNode
) input) and it is becoming a bit of a pain to deal with the limitations.We should unify these two modes and make sure that any index-join-specific optimizations are carried over and can be applied whenever we are doing an index join (e.g. #40748, #39471).
The text was updated successfully, but these errors were encountered: