[
https://issues.apache.org/jira/browse/CALCITE-2970?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17099483#comment-17099483
]
Julian Hyde commented on CALCITE-2970:
--------------------------------------
Can we have a new JIRA case focused just on PR 1884? The purpose seems to be to
switch RelBuilder to create rels of a different convention.
Reviewing the latest 1884:
* Struct's constructor should remain private (otherwise it will break every
time we add a new type of RelNode);
* RelBuilder's struct field should remain final;
* Change {{setRelFactoriesStruct}} to {{withFactories}} and make it create a
new {{RelBuilder}} with different config and factories but sharing the same
stack of RelNodes;
* need to document the new method in {{algebra.md}}.
(I acknowledge that having a new RelBuilder that shares state with a previous
RelBuilder is a bit fishy, but less fishy, I think, than mutating the config of
a RelBuilder. Because people expect that the only thing that mutates in a
RelBuilder is the stack of rels.)
This makes possible option C:
{code:java}
// Create a sort operator based on given convention
Convention toConvention = rel.getConvention();
RelBuilder builder = ...;
RelNode sort = builder.push(rel)
.adopt(toConvention)
.sort(toCollation)
.build();
{code}
where the {{RelBuilder.adopt}} method is as follows:
{code:java}
public final RelBuilder adopt(Convention convention) {
return convention.transformRelBuilder(this);
}
{code}
and a typical sub-class of {{Convention}} would implement the
{{transformRelBuilder}} method like this:
{code:java}
public RelBuilder transformRelBuilder(RelBuilder builder) {
return builder.withFactories(FOO_STRUCT);
}
{code}
> Performance issue when enabling abstract converter for EnumerableConvention
> ---------------------------------------------------------------------------
>
> Key: CALCITE-2970
> URL: https://issues.apache.org/jira/browse/CALCITE-2970
> Project: Calcite
> Issue Type: Bug
> Components: core
> Reporter: Haisheng Yuan
> Priority: Major
> Labels: pull-request-available
> Time Spent: 18h 20m
> Remaining Estimate: 0h
>
> If we enable the use of abstract converter for {{EnumerableConvention}}, by
> making {{useAbstractConvertersForConversion}} return true,
> {{JDBCTest.testJoinManyWay}} will not complete.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)