Kaijie Chen
Kaijie Chen
+1 支持 rsync://mirror.yandex.ru/mirrors/parabola/
Actually we can manually set active profiles in IDEA like this, So this change is not necessary, and the solution is limited to spark3 profile.
> I'm neutral for `activeByDefault`, but recommend to unify the profile style, e.g. > > ``` > spark-2.3 > spark-2.4 > spark-3.0 > spark-3.1 > spark-3.2 > ``` Agreed.
I would suggest borrowing the footer from [Apache Kvrocks](https://kvrocks.apache.org/).
And it's generally better to put website in a separate repo, such as [spark-website](https://github.com/apache/spark-website).
Thanks @adoroszlai for the comments. Actually I'm closing this for the same reason. This patch is very easy to make (by command), and we can wait for a good chance...
For example, is is possible to read stale `size` if context switch happens before line 66: https://github.com/apache/incubator-uniffle/blob/ec00ad91c72dc23b3e81b5d9abe815438b0f5bcf/server/src/main/java/org/apache/uniffle/server/buffer/ShuffleBuffer.java#L61-L67
> Reading long type (x86_64 word) is already atomic with or without synchronized. Here `size` means the sum of size of all blocks. It is possible that when you read...
> Like discussion at dev mail list, we will freeze the code and cut 0.6 version branch in September 15, we will not merge this pr before I cut 0.6...
What's the build command you used for building Uniffle against spark 3.2.1?