Github user jkleckner commented on the pull request:
https://github.com/apache/spark/pull/4780#issuecomment-76428126
>Are you packaging CA yourself, are you marking all the Spark deps as
provided? I see you're already using the ```userClassPathFirst``` settings.
Cases of sparkSubmit (fails => fastutil class not found exception) that
fail:
- Mark CA as provided
- Include CA in our assembly
- Include CA in our assembly and use
```spark.driver.userClassPathFirst=true``` and
```spark.executor.userClassPathFirst=true```
- Include CA and ```fastutil``` in our assembly
- Include CA and ```fastutil``` in our assembly and use
```userClassPath=true```
Cases that succeed:
- Include source for CA directly and rename package from ```clearspring```
to ```clearsprin```
- Include source for CA directly and rename package from ```clearspring```
to ```clearsprin``` and use ```userClassPath=true```
Note that these last two cases do cause ```fastutil``` to be included in
the assembly.
>(PS good news, I see that parquet-column's shading is set to only include
classes that are used from fastutil. That's great; there are only tens of
classes added. That's why they had to shade.)
I don't see a ```minimizeJar``` option with ```sbt-assembly``` which is
what we are using. I have to say, that would be nice since it is over 16MB...
So, I guess we will have to just copy/rename the CA package? Or should the
```userClassPathFirst``` work in this situation where spark is using one of the
classes?
---
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 is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
