Github user fhueske commented on the pull request:
https://github.com/apache/flink/pull/1544#issuecomment-176170046
Yes, I wanted to start a discussion about sticking with Scala as well.
I'll open a PR for FLINK-3225 very soon and can take care of translating
the nodes to Scala an
Github user twalthr commented on the pull request:
https://github.com/apache/flink/pull/1544#issuecomment-176150700
When I wrote my SQL prototype I also thought about using Java because
Calcite is written in Java, but since it is easy to use Java classes from Scala
and Scala has a muc
Github user ChengXiangLi commented on the pull request:
https://github.com/apache/flink/pull/1544#issuecomment-176144759
I didn't prefer Java or Scala, the reason why i writen this in Java is
that, this part logic would be translated from Calcite RelNodes(which is in
Java) and finally
Github user twalthr commented on the pull request:
https://github.com/apache/flink/pull/1544#issuecomment-176133567
@fhueske and @ChengXiangLi: Sorry but I have to discuss this PR again. Do
we really want to have parts of the Table API written in Java and some other
parts written in S
Github user ChengXiangLi closed the pull request at:
https://github.com/apache/flink/pull/1544
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature i
Github user fhueske commented on the pull request:
https://github.com/apache/flink/pull/1544#issuecomment-175506270
Ah, just saw that you committed already. Very good :-)
I think you can close this PR then.
---
If your project is set up for it, you can reply to this email and have
Github user fhueske commented on the pull request:
https://github.com/apache/flink/pull/1544#issuecomment-175506011
Hi, you cannot push to the Apache mirror, only to the writable Apache
repository: https://git-wip-us.apache.org/repos/asf/flink.git
I added the Apache repository
Github user ChengXiangLi commented on the pull request:
https://github.com/apache/flink/pull/1544#issuecomment-175365963
Just curious, about the merging the PR, the way i know is that, checkout
the tableOnCalcite, apply the patch, commit and push to flink repo, is this the
way how you
Github user ChengXiangLi commented on the pull request:
https://github.com/apache/flink/pull/1544#issuecomment-175360681
@fhueske , yes, i just remove the method in `FlinkRelNode` as java
interface only allow public method, just forgot to change the methods in
implemented FlinkRelNode
Github user fhueske commented on the pull request:
https://github.com/apache/flink/pull/1544#issuecomment-175119125
@ChengXiangLi, actually I proposed to make all methods other than
`translateToPlan()` private methods. If the Flink RelNodes directly return the
DataSet program, we do n
Github user twalthr commented on the pull request:
https://github.com/apache/flink/pull/1544#issuecomment-174936146
Thanks for the refactoring. IMO I think we can remove the "Flink" prefix if
we use DataSet/DataStream as prefix. Maybe it makes sense to add a
`TranslationContext` param
Github user fhueske commented on the pull request:
https://github.com/apache/flink/pull/1544#issuecomment-174908452
One more thing, we should rename the RelNodes to something like
`DataSetFlatMap` to indicate that these nodes can only be used for the DataSet
/ Batch translation. We wi
Github user fhueske commented on the pull request:
https://github.com/apache/flink/pull/1544#issuecomment-174900754
Just to minor comments from my side.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project doe
Github user fhueske commented on a diff in the pull request:
https://github.com/apache/flink/pull/1544#discussion_r50807436
--- Diff:
flink-libraries/flink-table/src/main/java/org/apache/flink/api/table/sql/calcite/FlinkRelNode.java
---
@@ -0,0 +1,31 @@
+/*
+ * Licensed to
Github user fhueske commented on a diff in the pull request:
https://github.com/apache/flink/pull/1544#discussion_r50807009
--- Diff:
flink-libraries/flink-table/src/main/java/org/apache/flink/api/table/sql/calcite/node/FlinkExchange.java
---
@@ -0,0 +1,62 @@
+/*
+ * Licen
Github user ChengXiangLi commented on the pull request:
https://github.com/apache/flink/pull/1544#issuecomment-174760809
Thanks for the comments, @fhueske and @twalthr . i've updated the PR.
---
If your project is set up for it, you can reply to this email and have your
reply appear
Github user fhueske commented on the pull request:
https://github.com/apache/flink/pull/1544#issuecomment-174518788
Yes, my initial idea was also to separate between 1) translating to the
correct job structure and 2) translation to actual DataSet program. So
basically the same motivat
Github user twalthr commented on the pull request:
https://github.com/apache/flink/pull/1544#issuecomment-174515248
Initially, I thought it makes sense to separate the code generation from
the optimization (and thus from Calcite). So that we could replace Calcite
easily at any time. B
Github user ChengXiangLi commented on the pull request:
https://github.com/apache/flink/pull/1544#issuecomment-174479220
Hi, @fhueske , What about `Project` and `Agggregate`? they are translated
to `FlinkMap` and `FlinkReduce` which includes `MapFunction` and
`GroupReduceFunction`. To
Github user fhueske commented on the pull request:
https://github.com/apache/flink/pull/1544#issuecomment-174471324
Thanks for this PR @ChengXiangLi.
This PR assumes that the code generation happens during the Calcite
translation process, because the interfaces return concrete
20 matches
Mail list logo
