Github user yaoshengzhe commented on the pull request:
https://github.com/apache/spark/pull/126#issuecomment-37492214
@tdas, based on what you said, it's fine to put logic into finalize if we
can make sure code in finalize is done in the right way and accept the fact
that fin
Github user yaoshengzhe commented on the pull request:
https://github.com/apache/spark/pull/126#issuecomment-37475144
@Tathagata, I am strongly disagree to put cleanup logic in finalize.
Finalizers are unpredictable, often dangerous, and generally unnecessary, e.g.
there is a severe
Github user yaoshengzhe commented on the pull request:
https://github.com/apache/spark/pull/120#issuecomment-37468388
From the code, it looks like this patch is only for renaming, is that
really important to use Yarn's jargon instead of master/worker ? I think
master/worker is
Github user yaoshengzhe commented on a diff in the pull request:
https://github.com/apache/spark/pull/120#discussion_r10540808
--- Diff:
yarn/common/src/main/scala/org/apache/spark/deploy/yarn/ClientArguments.scala
---
@@ -67,24 +67,39 @@ class ClientArguments(val args: Array
Github user yaoshengzhe commented on the pull request:
https://github.com/apache/spark/pull/95#issuecomment-36938296
@pwendell I agree what you saying. One more question, is that possible to
move all these string constants in some class ?
---
If your project is set up for it
Github user yaoshengzhe commented on the pull request:
https://github.com/apache/spark/pull/95#issuecomment-36935885
Hi Sandy,
What is the point to give a new name to "yarn-standalone" ? I think it
requires people to change their spark code build for yarn and c
Github user yaoshengzhe commented on the pull request:
https://github.com/apache/spark/pull/72#issuecomment-36929714
I think reduceByKeyToDriver is a better name and developer could easily
figure out the behavior of this operation. Instead, reduceByKeyLocally is a
little bit
