[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-09-28 Thread Hyukjin Kwon (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16185169#comment-16185169 ] Hyukjin Kwon commented on SPARK-21190: -- Will keep in mind and suggest to fix it or f

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-09-28 Thread Reynold Xin (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16185166#comment-16185166 ] Reynold Xin commented on SPARK-21190: - OK it would be great to have a better error me

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-09-28 Thread Hyukjin Kwon (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16185164#comment-16185164 ] Hyukjin Kwon commented on SPARK-21190: -- [~rxin], I suggested to disallow it [here|h

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-09-28 Thread Reynold Xin (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16185155#comment-16185155 ] Reynold Xin commented on SPARK-21190: - Where did we settle on 0-arg UDFs? I think we

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-09-23 Thread Reynold Xin (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16177886#comment-16177886 ] Reynold Xin commented on SPARK-21190: - Maybe create an umbrella ticket so it is easie

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-09-23 Thread Wenchen Fan (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16177885#comment-16177885 ] Wenchen Fan commented on SPARK-21190: - yea, let's do that in a separated ticket. > S

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-09-23 Thread Li Jin (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16177846#comment-16177846 ] Li Jin commented on SPARK-21190: [~cloud_fan], do we want to track other vectorized udf e

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-09-12 Thread Takuya Ueshin (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16162646#comment-16162646 ] Takuya Ueshin commented on SPARK-21190: --- [~icexelloss] Thank you for your suggestio

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-09-11 Thread Bryan Cutler (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16162205#comment-16162205 ] Bryan Cutler commented on SPARK-21190: -- Thanks [~icexelloss]. I definitely think co

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-09-08 Thread Li Jin (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16158845#comment-16158845 ] Li Jin commented on SPARK-21190: To [~bryanc]'s point, PR [#18659|https://github.com/apac

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-09-06 Thread Bryan Cutler (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16156082#comment-16156082 ] Bryan Cutler commented on SPARK-21190: -- I attached my PR because it had already been

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-09-06 Thread Apache Spark (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16156068#comment-16156068 ] Apache Spark commented on SPARK-21190: -- User 'BryanCutler' has created a pull reques

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-09-06 Thread Apache Spark (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16155271#comment-16155271 ] Apache Spark commented on SPARK-21190: -- User 'ueshin' has created a pull request for

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-09-06 Thread Takuya Ueshin (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16155270#comment-16155270 ] Takuya Ueshin commented on SPARK-21190: --- [~leif], [~bryanc] Thanks for the instruct

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-09-05 Thread Bryan Cutler (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16154361#comment-16154361 ] Bryan Cutler commented on SPARK-21190: -- Thanks [~ueshin], I think having an optional

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-09-05 Thread Leif Walsh (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16153986#comment-16153986 ] Leif Walsh commented on SPARK-21190: I think the size parameter is confusing: if a 1-

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-09-05 Thread Takuya Ueshin (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16153978#comment-16153978 ] Takuya Ueshin commented on SPARK-21190: --- [~leif] Thank you for your proposal. I'm s

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-09-05 Thread Takuya Ueshin (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16153955#comment-16153955 ] Takuya Ueshin commented on SPARK-21190: --- [~bryanc] Thank you for your suggestion. {

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-09-01 Thread Leif Walsh (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16151377#comment-16151377 ] Leif Walsh commented on SPARK-21190: You can also make a Series with no content and a

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-09-01 Thread Leif Walsh (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16151376#comment-16151376 ] Leif Walsh commented on SPARK-21190: Yep, that's totally a thing: {noformat}In [1]:

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-09-01 Thread Leif Walsh (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16151261#comment-16151261 ] Leif Walsh commented on SPARK-21190: I'm not 100% sure this is legal pandas but I thi

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-09-01 Thread Bryan Cutler (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16150924#comment-16150924 ] Bryan Cutler commented on SPARK-21190: -- I'm good with the API summary proposed by [~

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-08-28 Thread Wenchen Fan (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16143491#comment-16143491 ] Wenchen Fan commented on SPARK-21190: - hmmm, your proposal has a weird usage: users n

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-08-24 Thread Li Jin (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16140094#comment-16140094 ] Li Jin commented on SPARK-21190: [~ueshin], Got it. I'd actually prefer doing it this way

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-08-23 Thread Takuya Ueshin (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16139507#comment-16139507 ] Takuya Ueshin commented on SPARK-21190: --- [~icexelloss] We can know the length of in

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-08-23 Thread Li Jin (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16138340#comment-16138340 ] Li Jin commented on SPARK-21190: [~ueshin], thanks for the summary. +1 for this API. Alt

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-08-22 Thread Takuya Ueshin (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16137897#comment-16137897 ] Takuya Ueshin commented on SPARK-21190: --- Hi all, I'd like to summarize this discus

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-07-28 Thread Li Jin (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16105095#comment-16105095 ] Li Jin commented on SPARK-21190: I think the use case 2 of what [~rxin] proposed original

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-07-28 Thread Li Jin (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16105072#comment-16105072 ] Li Jin commented on SPARK-21190: [~cloud_fan], thanks for pointing out `ArrowColumnVector

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-07-26 Thread Bryan Cutler (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16102420#comment-16102420 ] Bryan Cutler commented on SPARK-21190: -- Hi [~icexelloss], yes I think there is defin

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-07-25 Thread Wenchen Fan (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16101014#comment-16101014 ] Wenchen Fan commented on SPARK-21190: - I think (2) is already done by {{ArrowColumnVe

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-07-25 Thread Li Jin (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16100712#comment-16100712 ] Li Jin commented on SPARK-21190: [~bryanc], I have looked at your PR at https://github.c

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-07-25 Thread Li Jin (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16100582#comment-16100582 ] Li Jin commented on SPARK-21190: I have created this PR for the groupby().apply() use cas

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-07-11 Thread Bryan Cutler (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16083018#comment-16083018 ] Bryan Cutler commented on SPARK-21190: -- [~cloud_fan] yes, I know not every function

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-07-10 Thread Wenchen Fan (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16079954#comment-16079954 ] Wenchen Fan commented on SPARK-21190: - [~bryanc] You example works because Python is

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-07-07 Thread Bryan Cutler (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16078837#comment-16078837 ] Bryan Cutler commented on SPARK-21190: -- [~rxin] I was talking about 2 different thin

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-07-06 Thread Reynold Xin (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16077484#comment-16077484 ] Reynold Xin commented on SPARK-21190: - [~bryanc] Sorry I don't think it makes sense t

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-07-06 Thread Bryan Cutler (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16077390#comment-16077390 ] Bryan Cutler commented on SPARK-21190: -- This is a great discussion so far and I wou

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-07-05 Thread Leif Walsh (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16074857#comment-16074857 ] Leif Walsh commented on SPARK-21190: If the user specifies an int return type but pro

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-07-05 Thread Li Jin (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16074848#comment-16074848 ] Li Jin commented on SPARK-21190: > I have 2 thoughts: > 1. How should we handle null valu

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-07-03 Thread Leif Walsh (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16072429#comment-16072429 ] Leif Walsh commented on SPARK-21190: I believe we could also compute window indexes w

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-07-03 Thread Leif Walsh (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16072424#comment-16072424 ] Leif Walsh commented on SPARK-21190: I figure we could address that by using shared m

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-07-03 Thread Wenchen Fan (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16072019#comment-16072019 ] Wenchen Fan commented on SPARK-21190: - > I think we can get away with doing windowing

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-06-30 Thread Wenchen Fan (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16071002#comment-16071002 ] Wenchen Fan commented on SPARK-21190: - Thanks for your proposal! I have 2 thoughts:

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-06-30 Thread Leif Walsh (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16070809#comment-16070809 ] Leif Walsh commented on SPARK-21190: I think we can get away with doing windowing (de

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-06-30 Thread Li Jin (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16070618#comment-16070618 ] Li Jin commented on SPARK-21190: I have some APIs design written down here: Here is how t

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-06-30 Thread Wenchen Fan (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16069665#comment-16069665 ] Wenchen Fan commented on SPARK-21190: - For aggregate, I think it makes more sense to

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-06-29 Thread Reynold Xin (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16069490#comment-16069490 ] Reynold Xin commented on SPARK-21190: - That makes a lot of sense. So to design APIs s

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-06-29 Thread Leif Walsh (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16069269#comment-16069269 ] Leif Walsh commented on SPARK-21190: I agree with [~icexelloss] that we should aim to

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-06-26 Thread Li Jin (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16063873#comment-16063873 ] Li Jin commented on SPARK-21190: [~r...@databricks.com], The use case of seeing entire p

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-06-26 Thread Reynold Xin (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16063515#comment-16063515 ] Reynold Xin commented on SPARK-21190: - [~icexelloss] Thanks. Your proposal brings up

[jira] [Commented] (SPARK-21190) SPIP: Vectorized UDFs in Python

2017-06-26 Thread Li Jin (JIRA)
[ https://issues.apache.org/jira/browse/SPARK-21190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16063103#comment-16063103 ] Li Jin commented on SPARK-21190: Very excited to see this. I created https://issues.apa