Uh oh!
There was an error while loading. Please reload this page.
[SPARK-13067][SQL] workaround for a weird scala reflection problem - #10970
[SPARK-13067][SQL] workaround for a weird scala reflection problem#10970cloud-fan wants to merge 1 commit into
Conversation
cloud-fan
commented
Jan 28, 2016
retest this please |
SparkQA
commented
Jan 29, 2016
Test build #50316 has finished for PR 10970 at commit
|
cloud-fan
commented
Jan 29, 2016
retest this please |
SparkQA
commented
Jan 29, 2016
Test build #50320 has finished for PR 10970 at commit
|
SparkQA
commented
Jan 29, 2016
Test build #2472 has finished for PR 10970 at commit
|
rxin
commented
Jan 29, 2016
"。" replace with "." ? |
davies
commented
Jan 29, 2016
Tested it locally, test passed. LGTM, I'm merging this into master, thanks! (fixed the 。) |
rxin
commented
Jan 29, 2016
@cloud-fan is this fixing a bug? If yes, please add a test somewhere. |
marmbrus
commented
Jan 29, 2016
Existing tests were failing locally, but not on Jenkins. We are not sure why. |
rxin
commented
Jan 29, 2016
OK got it. @cloud-fan Please add that to the pull request description next time. |
cloud-fan
commented
Jan 29, 2016
Sorry I thought the JIRA description tells it, will add it to PR description next time. |
rxin
commented
Jan 29, 2016
Yea the pull request description gets merged as commit description, and is the only thing that sticks around forever. |
atronchi
commented
Sep 2, 2016
How can we workaround this problem when using 1.6.1? |
atronchi
commented
Sep 6, 2016
This solution does not seem to work for datasets on tables using complex types. e.g. Array[Int] still throws the exception |
cloud-fan
commented
Sep 7, 2016
@atronchi can you create a JIRA and put the code that can reproduce the bug? thanks! |
atronchi
commented
Sep 7, 2016
The solution mentioned in [SPARK-17424] by @rdblue fixes this issue. |
A simple workaround to avoid getting parameter types when convert a logical plan to json. Note that this bug only exists locally, but ok on jenkins, not sure why...