Bryce Mecum

So what would you say the check would help you with in this case? The idea of adding fork prevention in FlightClient is that the user would get a warning...

I think maybe for the scope of this ticket, adding a PID-check just to FlightClient and improving the docs would be good start. If we find others are interested in...

Hi @pitrou, I did some testing so we're clear what actually breaks and how. I've pushed four separate reproducers to a branch on my fork at https://github.com/amoeba/arrow/tree/gh-38617/gh-38617 (see README in...

Yeah, I think that's the situation we want to cover here. I was originally thinking this could just be built into FlightClient and not the entire DLL/module: Save the PID...

> When you say "Forking then using a FlightClient from the child process" above, was the FlightClient created in the parent before forking or in the child? Yeah, I should've...

Thanks for filing this @robomics. I'm not familiar with the `create_library_symlinks` function yet but my initial take is that we might be better off giving the user a clear warning...

Really cool, @mbjones, I hadn't heard of a closure table before. I think having one or more O(1) APIs around version chains would be very helpful across current and future...

Since a version chain can be infinitely long, it seems to me like we'd want full pagination support. This'd mean having a `start` or `offset` argument to pair with `count`...

We discussed this on our Sep 10, 2020 dev call and continued in a break out afterwards for a bit. I'm commenting here to bring some of that discussion back...

I put a twin issue for this over on https://github.com/DataONEorg/d1_cn_index_processor/issues/14 too to track changes across the repos.