Christiano Haesbaert

IMHO this is far from ready, we didn't really review much, we just got it to work.

> I see this with `EIO_BACKEND=luv`. I suspect this is because we use a separate Luv event loop for each domain, but cohttp is sharing the socket between domains. Sockets...

That works, but there are many things involved, multiple sockets with reuseaddr/port work in a push manner. The kernel round robins the n XT connection, which means that your next...

That's nice ! I can promise to have a look but I need some time, same for the second PR.

yes please, don't default to max, I think forcing the user to pass the value was fine.

This is not the same as an unbound socket, I'm re-opening.

> At a high level, do we have a sense of what modern consumers of signals actually expect? The lowlevel interface is extremely difficult to get right in the presence...

SIGTERM for multiple domain cleanup is a kinda common thing. Not supporting SIGHUP seems unecessary, I don't see why we should remove this from the user. Cancelling fibers with SIGINT...

> SIGTERM for multiple domain cleanup is a kinda common thing. Not supporting SIGHUP seems unecessary, I don't see why we should remove this from the user. Cancelling fibers with...

I think the debugging is imprecise, this is pretty much just after the first deference of `s`: `s->unique_id = fresh_domain_unique_id();` `s` is probably screwed up, which means `&d->interruptor` is bad....