[neomutt-devel] To fork or not to fork?

[Guyzmo]

On Wed, Jan 25, 2017 at 09:45:31PM +0100, Guyzmo wrote:
>     * [libraries] should more functions be replaced by moving to gnulib or libuv?

I think the answer to that question shall be yes, without thinking
twice. We need to cut the fat, and get rid of code that has been crafted
for mutt that exists in libraries that had more scrutinity and is well
tested,

>     * "libuv" → depends on cmake

it'd be great to dig a bit further in this discussion, and see what we'd earn by
switching to libuv vs gnulib vs existing.

What I find interesting with libuv is that it's an asynchronous I/O
library, so it would make it possible to keep neomutt responsive while
doing I/O operations (like loading the mails, doing network operations
etc.).

As libuv uses threads within a threadpool to interact, it should be a
no brainer to integrate. Because neomutt has an ncurse eventloop, it
should be reasonably possible to take advantage of the async features.
Otherwise, it shouldn't be impossible to emulate the synchronous logic.

https://nikhilm.github.io/uvbook/

I don't know much about what's included in gnulib, if someone could
lecture us about that! What would it bring to a codebase?

Cheers,

-- 
Guyzmo