Github user srowen commented on the issue:
https://github.com/apache/spark/pull/14524
Let's back off this for now; see instead
https://github.com/apache/spark/pull/14826 for just the 'fix'
---
If your project is set up for it, you can reply to this email and have your
reply appear on
Github user srowen commented on the issue:
https://github.com/apache/spark/pull/14524
I probably need some oversight here since I'm not familiar with pyspark in
this regard, but it looks like PythonMLLibAPI treated seeds a little
differently in a few cases which meant they couldn't be
Github user BryanCutler commented on the issue:
https://github.com/apache/spark/pull/14524
Just to clarify what we are talking about on the Python side, pyspark-mllib
given `seed=None` will generate a random seed based on system time, while
pyspark-ml given `seed=None` will use a fixe
Github user srowen commented on the issue:
https://github.com/apache/spark/pull/14524
Yes, isn't that why it's possible to fix a seed? I can understand an
argument that the default should be not-random, but, every API I've ever seen
(including Spark's) defaults to a random seed.
Github user mengxr commented on the issue:
https://github.com/apache/spark/pull/14524
Sorry for late response! I'm against this change since it introduces
indeterministic behavior and makes applications hard to debug. For example, I
want to cross validate some estimator that accepts a
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/14524
Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/64193/
Test FAILed.
---
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/14524
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
e
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14524
**[Test build #64193 has
finished](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/64193/consoleFull)**
for PR 14524 at commit
[`ceebf7c`](https://github.com/apache/spark/commit/
Github user felixcheung commented on the issue:
https://github.com/apache/spark/pull/14524
+ @yanboliang for R `spark.gaussianMixture`
---
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 fea
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14524
**[Test build #64193 has
started](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/64193/consoleFull)**
for PR 14524 at commit
[`ceebf7c`](https://github.com/apache/spark/commit/c
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/14524
Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/64165/
Test FAILed.
---
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/14524
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
e
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14524
**[Test build #64165 has
finished](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/64165/consoleFull)**
for PR 14524 at commit
[`9d1ea81`](https://github.com/apache/spark/commit/
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14524
**[Test build #64165 has
started](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/64165/consoleFull)**
for PR 14524 at commit
[`9d1ea81`](https://github.com/apache/spark/commit/9
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14524
**[Test build #64064 has
finished](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/64064/consoleFull)**
for PR 14524 at commit
[`26fa2e0`](https://github.com/apache/spark/commit/
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/14524
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
e
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/14524
Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/64064/
Test FAILed.
---
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14524
**[Test build #64064 has
started](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/64064/consoleFull)**
for PR 14524 at commit
[`26fa2e0`](https://github.com/apache/spark/commit/2
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14524
**[Test build #64060 has
finished](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/64060/consoleFull)**
for PR 14524 at commit
[`ceb5050`](https://github.com/apache/spark/commit/
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/14524
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
e
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/14524
Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/64060/
Test FAILed.
---
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14524
**[Test build #64060 has
started](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/64060/consoleFull)**
for PR 14524 at commit
[`ceb5050`](https://github.com/apache/spark/commit/c
Github user srowen commented on the issue:
https://github.com/apache/spark/pull/14524
Jenkins retest this please
---
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 wishe
Github user srowen commented on the issue:
https://github.com/apache/spark/pull/14524
OK, this sounds like it could be a good change. I'd like to ask @jkbradley
again in case there was a strong reason I'm missing to fix the seed by default.
Also CC @davies if possible to consider the
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/14524
Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/63781/
Test PASSed.
---
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/14524
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
e
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14524
**[Test build #63781 has
finished](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/63781/consoleFull)**
for PR 14524 at commit
[`ceb5050`](https://github.com/apache/spark/commit/
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14524
**[Test build #63781 has
started](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/63781/consoleFull)**
for PR 14524 at commit
[`ceb5050`](https://github.com/apache/spark/commit/c
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14524
**[Test build #63743 has
finished](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/63743/consoleFull)**
for PR 14524 at commit
[`af2bdb0`](https://github.com/apache/spark/commit/
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/14524
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
e
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/14524
Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/63743/
Test FAILed.
---
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14524
**[Test build #63743 has
started](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/63743/consoleFull)**
for PR 14524 at commit
[`af2bdb0`](https://github.com/apache/spark/commit/a
Github user srowen commented on the issue:
https://github.com/apache/spark/pull/14524
@mengxr do you happen to have an opinion on this?
---
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 fe
Github user BryanCutler commented on the issue:
https://github.com/apache/spark/pull/14524
+1 for changing this. If left as is, JIRAs like this one will be a
recurring theme. A lot of users would probably not even notice, then wind up
getting degraded results and not know why.
---
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/14524
Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/63386/
Test FAILed.
---
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14524
**[Test build #63386 has
finished](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/63386/consoleFull)**
for PR 14524 at commit
[`da0d315`](https://github.com/apache/spark/commit/
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/14524
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
e
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14524
**[Test build #63386 has
started](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/63386/consoleFull)**
for PR 14524 at commit
[`da0d315`](https://github.com/apache/spark/commit/d
Github user srowen commented on the issue:
https://github.com/apache/spark/pull/14524
Yes, it makes sense to try to fix seeds in tests for reproducibility, and
most of the code does that already (hence tests pass) but right now it doesn't
seem possible to get random behavior unless yo
Github user holdenk commented on the issue:
https://github.com/apache/spark/pull/14524
I share some of @MLnick's concern we've looked at similar changes with
using system random seeds for the defaults in other places and had difficult
with making reproducible tests - in general we use
Github user MLnick commented on the issue:
https://github.com/apache/spark/pull/14524
I tend to agree the default should be a new random seed, and it is more
consistent with other libs. But @jkbradley seemed to explicitly want things to
default to reproducible behavior in
[SPARK-7678
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/14524
Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/63325/
Test PASSed.
---
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/14524
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
e
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14524
**[Test build #63325 has
finished](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/63325/consoleFull)**
for PR 14524 at commit
[`fbc93eb`](https://github.com/apache/spark/commit/
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14524
**[Test build #63325 has
started](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/63325/consoleFull)**
for PR 14524 at commit
[`fbc93eb`](https://github.com/apache/spark/commit/f
Github user srowen commented on the issue:
https://github.com/apache/spark/pull/14524
Want to check with @jkbradley on this before proceeding, but seems to pass
tests
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If yo
46 matches
Mail list logo