-
Notifications
You must be signed in to change notification settings - Fork 1
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
The Quest for Performance: smart Rete layouts #5
Comments
By Gabor Bergmann on Feb 13, 2013 13:26 New refinement requiring functional dependency analysis:\
|
By Gabor Bergmann on Mar 12, 2013 19:37
|
By Gabor Bergmann on Mar 13, 2013 10:19 More ideas:
|
By Gabor Bergmann on Apr 19, 2013 04:39 Functional dependencies again: prefer the join candidate where the operands have a smaller closure (the sum of their closure is smaller?). Motivation: if X->Y->Z, joining Y and Z first and then joining X against (YZ) is more efficient, as there are probably fewer variations of Y and Z (not more than variations of X). |
By Gabor Bergmann on May 10, 2013 08:22
|
By Gabor Bergmann on May 10, 2013 14:45 In commit a8dd786, Added eager trimming of variables - no planning ahead yet. Let's hope for good results. |
By Istvan Rath on May 20, 2013 04:35 I'm setting version to "unspecified" as we currently do not (firmly) know when these future ideas will be addressed. |
By Gabor Bergmann on Jun 05, 2014 05:51 The query planner should recognize deferred constraints that depend on no variables, and treat them as enumerable. Examples:
|
By Gabor Bergmann on Jun 02, 2015 10:14 Foreseen improvements to support the decision making of the query planner:
|
By Abel Hegedus on Apr 19, 2016 05:24 Updating to correct milestone. |
By Gabor Bergmann on May 09, 2016 11:32
|
By Gabor Bergmann on May 09, 2016 11:37
...and change TupleMask#combine so that it just returns unmasked for empty complementer masks - thus filter-only joins will reuse tuples, should be nice memory win. |
By Gabor Bergmann on May 27, 2016 03:54
Improve the currently implemented trimming so that it results in the same order of indexes every time - improves reuse of trimmer nodes and their descendants |
By Gabor Bergmann on Jul 06, 2018 03:55 Refactor the notion of 'Production' node so that it is not a node type but rather a role that pretty much any Supplier node may take. Foreseen benefits: (a) saves memory when there is no need to enforce uniqueness (single body, no projection at the end), and a join or trimmer node can represent the match set of the pattern; (b) better node sharing among isomorphic patterns. Be careful with recursion, with traceability and with how the result appears in Rete visualizer. |
| --- | --- |
| Bugzilla Link | 398763 |
| Status | NEW |
| Importance | P3 normal |
| Reported | Jan 22, 2013 11:02 EDT |
| Modified | Jul 06, 2018 03:55 EDT |
| Version | oldinquery |
| Depends on | 399667, 493140, 493256 |
| Reporter | Gabor Bergmann |
Description
Cloned from: 326: The Quest for Performance: smart Rete layouts
http://github.com/ujhelyiz/EMF-IncQuery/issues/issue/326
Implement various ideas for building a more clever Rete structure:
The text was updated successfully, but these errors were encountered: