Steve Loughran

If you can address those little nits and and are confident that the errors reported go away, then yes.

I don't normally go near hdfs, so had missed this. It also means: I don't review their patches. I think this should target 3.4.2 and we focus on getting that...

Raising an java.io.EOFException when passing down an offset/range > file length is what is required, so all is good. s3a and abfs support the openFileWithOptions so they can save a...

> , it would be good to create a VectoredInputStream which takes the actual DataInputStream as input and then all the Object stores like abfs, s3 and allyun extending this...

have you tested this through parquet 1.14.1 yet? it supports vector io -just turn it on! I'd love to see what speedups you get

+test this on google gcs through the connector and see what failure exception it raises; have it translate to the same ioe

@diljotgrewal upgrading an aws SDK MUST be done as a standalone action. we have a whole section of the testing doc [that covers this](https://hadoop.apache.org/docs/stable/hadoop-aws/tools/hadoop-aws/testing.html#Qualifying_an_AWS_SDK_Update). please create the jira and follow...

FYI I'm working on a document covering how we can modify the output stream, rename code and more to deliver this more broadly. Will share soon

@shameersss1 exactly. which means that fs.create(path, false).close() is equivalent to other filesystems, though ``` st = fs.create(path, false) st.write("I think I have exclusive access") st.close() ``` is not quite the...

@diljotgrewal except in the special case of "massive merge conflict requiring a rebase", can you just use merge commits once we are in the review phase. Github lets me review...