Github user felixcheung commented on the issue:
https://github.com/apache/spark/pull/19027
Sure - I think there are a number of different situations reported in the
JIRA that could be separated into different fixes.
Let me know what I can help with!
---
If your project is
Github user HyukjinKwon commented on the issue:
https://github.com/apache/spark/pull/19027
Merged to master.
---
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
Github user HyukjinKwon commented on the issue:
https://github.com/apache/spark/pull/19027
Will merge this one BTW. Sounds we are fine.
---
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
Github user HyukjinKwon commented on the issue:
https://github.com/apache/spark/pull/19027
That's fine, @ueshin and @felixcheung. Adding few tests with `numpy` type
might be an extra bit and (possibly) unrelated vs it's easy to add a test and
might be a (possibly) common case users
Github user ueshin commented on the issue:
https://github.com/apache/spark/pull/19027
@felixcheung I'm sorry if I'm missing something but it sounds like it's a
different problem from this pr?
---
If your project is set up for it, you can reply to this email and have your
reply
Github user felixcheung commented on the issue:
https://github.com/apache/spark/pull/19027
It's not specific to it, but fairly common when people are calling numpy in
UDF and returning its scalar type as-is. These scalar "looks" like Python
native types (numpy.float_ vs float).
Github user ueshin commented on the issue:
https://github.com/apache/spark/pull/19027
LGTM.
Btw, I'm just curious why we need tests with `numpy` here.
---
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
Github user HyukjinKwon commented on the issue:
https://github.com/apache/spark/pull/19027
Will probably take a look through the problem in the near future including
hard dependencies and etc. I took a quick look but I think I need more time but
yes it looks appearently vaild point.
Github user felixcheung commented on the issue:
https://github.com/apache/spark/pull/19027
I'm ok without the test since this is unlikely to break in the future. We
do have tests that depends on (optionally) numpy (and Arrow) - seems like we
should be able to take on dependencies
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/19027
Merged build finished. Test PASSed.
---
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
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/19027
Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/81056/
Test PASSed.
---
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/19027
**[Test build #81056 has
finished](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/81056/testReport)**
for PR 19027 at commit
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/19027
**[Test build #81056 has
started](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/81056/testReport)**
for PR 19027 at commit
Github user HyukjinKwon commented on the issue:
https://github.com/apache/spark/pull/19027
Oops, looks I need to check if numpy is available. Let me rather take this
one out here as I am trying to whitelist `basestring` if you don't mind. I
tested it with numpy in my local for your
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/19027
Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/81053/
Test FAILed.
---
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/19027
**[Test build #81053 has
finished](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/81053/testReport)**
for PR 19027 at commit
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/19027
Merged build finished. Test FAILed.
---
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
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/19027
**[Test build #81053 has
started](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/81053/testReport)**
for PR 19027 at commit
Github user HyukjinKwon commented on the issue:
https://github.com/apache/spark/pull/19027
Thanks @felixcheung and @holdenk. I just added a simple test with
numpy.float.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well.
Github user holdenk commented on the issue:
https://github.com/apache/spark/pull/19027
I like this approach @HyukjinKwon :D!
---
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
Github user felixcheung commented on the issue:
https://github.com/apache/spark/pull/19027
Cool looks to me like a very reasonable fix.
Could we perhaps add a test for numpy.bool_ or numpy.float_ (that it should
fail)?
---
If your project is set up for it, you can reply to this
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/19027
Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/81030/
Test PASSed.
---
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/19027
Merged build finished. Test PASSed.
---
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
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/19027
**[Test build #81030 has
finished](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/81030/testReport)**
for PR 19027 at commit
Github user HyukjinKwon commented on the issue:
https://github.com/apache/spark/pull/19027
cc @zero323, @rdblue, @nchammas, @holdenk, @ueshin and @felixcheung. Could
you take a look please? I think it is a small fix but the advantage is quite
large.
---
If your project is set up
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/19027
**[Test build #81030 has
started](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/81030/testReport)**
for PR 19027 at commit
26 matches
Mail list logo