Re: nanomsg binding?

[Petr Štetiar <ynezz@...>]

Pierre Chapuis <catwell@archlinux.us> [2013-11-11 16:05:54]:

> Currently thinking about what exactly I will say at the Workshop,
> and this topic in particular. I think it's a bit sad to see such
> wheel reinvention.

Yep, but what choices do we have here? Your luajit-nanomsg needs FFI,
lua-nanomsg is borken and to fix it properly, one would have to learn
LuaNativeObjects. And instead of wasting time mastering LuaNativeObjects (I'm
trying being pragmatic, so this is knowledge which I wouldn't reuse anywhere
else so waste of time) I would rather invest more time and write plain C
bindings from scratch.

And NanoMsg is so clean in my eyes, that I find lua-nanomsg quite bloated just
for it's bindings (no offense here, I appreciate Neopallium's work) :-)

> I can understand that different bindings can have different goals,
> but at least we should try to agree on a reasonable interface and
> make them interchangeable.

Good point and BTW I like the API of luajit-nanomsg.

> Regarding nanomsg, I had written a binding for LuaJIT in its early
> days and found out somebody else had as well. So we merged and the
> binding is now under the official nanomsg project on GitHub:
> https://github.com/nanomsg/luajit-nanomsg

I would use it, but FFI...

> I really don't think a library still very unstable should have that
> many bindings.

Unstable in which terms? I find it pretty usable in my use case (tm) so far.

-- ynezz