Forced stat collection fix #37
Closed
Conversation
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
the path_utils.c external functions. I want to rewrite clause manipulating subsystem and this header make the code more clear.
For each new branch we will rename aqo_master.patch
plan creating step. It must greatly reduce the AQO core patch.
It allows extensions to store an planning info for using during (or after) execution. Introduce the AQOPlanNode structure. Move AQO-related fields of the Plan structure in this structure. Add this structure as an element to the Plan->private field. We allow copying of AQOPlanNode and disallow _read... and _out... functions because it have sense and it needs to teach the Core to serialize/deserialize RestrictInfo and all descendants of this struct. Also, for simplifying the Core patch: 1. replace parallel_workers field with parallel_divisor 2. move IsParallelTuplesProcessing() from the Core into the AQO code. This commit changes the Core patch.
We will add into the head of each plan fss hash of this plan tree. it would allow us to account this as subplan in the AQO feature space of overall plan without plunging into this subplan.
does a copy od each clause and changes SubPlan clauses with the hash value corresponding to this SubPlan. So we get in account subplans. On the learning stage we plunge into a subplan (for learning on underlying nodes) but don't get into account its clauses, relids and selectivities. Regression tests on this feature added. Still in the unsupported.sql.
It was required to change ExplainOneNode_hook code and parameters. Now, we can show AQO prediction in EXPLAIN ONLY mode. The Core patch is changed.
Previously we suppressed such plans because 'never executed' node is a kind of NULL value. We don't really know how many tuples (and work) it require in the situation, when it will be executed even once. Current commit is attempt to teach on such plans, may be partly. We allow teaching on all nodes, that were executed. If node was 'never executed' and no any information we have in ML-knowledge base for this feaure subspace - we add knowledge that this node can produce one tuple.
We could use it for cleaning old ML-data for previously dropped tables.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
No description provided.