Ziheng Wang

What is the status with that PR? I don't see any comments on it or any reviewers assigned.

Some comments in the code probably has to be changed. But let's first agree on the actual interface and functionality. @pitrou @westonpace

OK I think I have passed CI now finally, please comment on the interface @pitrou @westonpace

- I think Fragment should be immutable. Instead of set_bounds can we have a Slice method like we have in python that returns a new Fragment (admittedly, since Fragment is...

This has been addressed

Checks that fail are because posix_fadvise doesn't work on windows/mac.... I have gated the code

Addressed all documentation related comments.

Actually I misunderstood the requirements. I think that synchronous writing is not required in this use case, and it is okay to use O_DIRECT without O_SYNC, to use the SSD...

[experiments.zip](https://github.com/apache/arrow/files/9339348/experiments.zip) Uploaded some code that can be used for benchmarking.

Let's review our options here: 1) O_DIRECT with O_SYNC. Horribly slow with the same problems with O_DIRECT, not worth it if you don't want to persist every write. 2) O_DIRECT...