Rudi Grinberg

I almost forgot to report the results of my experiment: `org-cycle` and `org-global-cycle` seem to be adequate substitutes to outshine's versions of these functions. I did not notice any differences...

> cram tests? These would be quite handy to understand & document the quirks of coqdoc, coqdep, coqc, etc.

I worry that this isn't flexible enough perhaps. Shouldn't this be controllable at the field label level?

+1 on the above. Most vim plugins don't even make sense in VsVim anyway. The only plugins that make sense are "editing" plugins. Those should be simple enough to implement...

Agreed. This is all caused by the fact that the server piece is heavily tied to conduit. Imo, we should have a simpler server layer that's based on `Lwt_io.establish_server` and...

Yes, this will be fixed. `Connection.t` is a useless value and will be removed. The individual backends will provide something useful in its stead.

Raising and catching an exception is surely much slower than a conditional. But I'm not sure if it's the same change semantically. ``` if Lwt.is_closed t then foo t ```...

Indeed. In particular, we'd prefer to kick out logging from the core of the system. It should be possible to add your own logging in a modular way at the...

Re-opening because this needs to be done for Async as well.

@anuragsoni i'm looking at cohttp_async/server.mli and i don't see an api for creating a server without conduit. Am I missing something?