Benjamin Fry

Thank you for this background. It’s great. For threading related things, I was considering serializing access to the allocator (if possible). Do you think this would be safe?

> So you could have the Postgres backend process thread allocating memory while some backround Rust thread is too. That’s an excellent point. I hadn’t considered it. The really big...

Cool that we thought of doing the same thing! I had a couple goals for this project, they done necessarily line up with yours, but maybe they do. It sounds...

> I have the feeling most rust panics would just be ERRORs, because unwinding should clean up any bad state, but I could be missing some potential problems. I read...

Following up on this, @jeff-davis we've implemented setjmp/longjmp handling in the library. It requires wrapping calls out to PG in a function we called `guard_pg`, which through testing appears to...

Yes, that would be very cool. I don’t have any strict plans for this library other than making it easy to build Postgres extensions in Rust. There’s still a lot...

looks that way. It looks like they're building directly on top of the FFI, rather than trying to use generic wrappers like `pg-extend-rs` is exporting. I didn't find `rgffi` when...

It looks like we need a way to define `_PG_output_plugin_init` similar to how we do `pg_magic` macro. In addition to that, it looks like a lot of C ffi functions...

I don't think we've ever tried to support 9.6. I think we could, someone would need to put in some amount of effort for it...

I think that would a great addition. It would probably need to be optional, as it would require the postgis extension to be installed. I think that could be a...