[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Gneuralnetwork] [PATCH] Compile gneuralnetwork as library
From: |
Nala Ginrut |
Subject: |
Re: [Gneuralnetwork] [PATCH] Compile gneuralnetwork as library |
Date: |
Sat, 30 Apr 2016 12:40:54 +0800 |
Hi Aljosha!
On Sun, 2016-04-24 at 17:01 +0200, Aljosha Papsch wrote:
>
> I could write more arguments in favor of Guile but I'd rather come to
> the practical side of things: How can Guile be integrated in
> gneuralnetwork?
>
I think FFI is the preferred way. Of course, there's problem in current
Guile FFI, for an instance, we don't save errno immediately after
calling C function, so errno could be changed in certain extreme
situation.
I've formated a patch for it months ago, but it's still suspended to
apply:
http://bug-guile.gnu.narkive.com/K6lxUWOI/bug-18592-ffi-should-have-por
table-access-to-errno
But anyway, FFI is still the preferred way to go. We'll fix the problem
in the future.
> steps:
>
> * Write Guile bindings using FFI
> * Write Scheme programs equivalent to the .script files
>
Agreed.
> While doing this, it should be possible to keep parser.c, so .script
> files can still be used. They could even be used when Guile support is
> proper. What is your stance on this?
>
> Best regards
> Aljosha
>
> [0] https://gnu.org/software/guile/
> [1] http://www.draketo.de/english/wisp/
> [2]
> https://gnu.org/software/guile/manual/html_node/Foreign-Function-Interface.html
Actually, once we support Guile, every language frontend implemented on
Guile could take advantage of the gneural_network binding. We just need
to maintain one binding for all. I'm working on guile-lua-rebirth, so
gneural_network could use Lua for scripting naturally in the future.