[Haskell-cafe] Re: ANNOUNCE: X Haskell Bindings 0.3

Antoine Latter aslatter at gmail.com
Sun Jun 28 14:12:35 EDT 2009


On Sun, Jun 28, 2009 at 12:52 PM, Antoine Latter<aslatter at gmail.com> wrote:
> I'd like to announce the 0.3.* series release of the X Haskell
> Bindings.  This release, like the prior 0.2.* series focuses on making
> the API prettier. This release is based on the XCB protocol
> descriptions version 1.5
>

I'd also like to share my thoughts on where the XHB project is, and
where I see it going.

As an experiment to show that an X client library can be written in
Haskell, I feel like XHB has succeeded.

But to be truly useful as a library, I feel like it needs to be able
to inter-operate with xlib and libxcb, which XHB currently does not.

Conceptually, XHB can be thought of as two parts:

 - Haskell data types generated from the X protocol description, which
know how to serilaize/deserialize themselves to and from byte streams.

 - A runtime which knows how to open a connection to the X server, and
knows how to peek and poke stuff onto the connection to the X server
(this is where all of the networking and thread wrangling code lives).

So it seems to me that the path forward is to keep as much of the
first part as possible, and then replace the second part with haskell
bindings to XCB.

I've started the binding to libxcb here:
http://community.haskell.org/~aslatter/code/xcb-ffi/


There are still a few unknowns in my mind, such as
 - I've never made Haskell bindings to a multi-threaded C App. Is
there anything strange I have to worry about?

 - I have no idea how I'll map event and error codes to the function
to decode the bytes into a Haskell type. In XHB all of the information
needed lives in Haskell datatypes - once I start moving over to being
a wrapper around XCB, the extensions cache is held in libxcb and isn't
exposed publicly.

But presumably these problems have solutions.

Antoine


More information about the Haskell-Cafe mailing list