unofficial mirror of guile-user@gnu.org 
 help / color / mirror / Atom feed
From: Matt Wette <matt.wette@gmail.com>
To: guile-user@gnu.org
Subject: Re: ffi helper
Date: Mon, 13 Mar 2017 17:53:59 -0700	[thread overview]
Message-ID: <95B6EEEF-5278-44C8-B640-647A16AEFD1F@gmail.com> (raw)
In-Reply-To: <96434CF4-E3F9-4410-82B6-9EB8DEE130B9@gmail.com>


> On Mar 8, 2017, at 6:06 PM, Matt Wette <matt.wette@gmail.com> wrote:
> 
> I’m now working on a FFI helper based on the nyacc C99 parser.   

So, like, I think what might be useful is something that will convert a dot-h file to a spec file, with certain assumptions about function signatures (e.g., if passing a pointer to int, then the scheme implementation is a (system foreign) int*, or maybe a boxed integer).  

Then the spec is compiled, in a machine-dependent manner to deal with type sizes and alignment, to a .go file which provides scheme-callable functions.  I’m not sure the output of the dot-H reader can be machine independent, given all the spaghetti code that typically exists in /usr/include.  It may be worth a try, later.

Also, the module should support structures.  The bytestructures package looks promising here.

Enums should be provided also.

CPP constants (aka #define’s) should be supported.

Do CPP macros with arguments need to be supported?   This is not just converting to functions, because CPP macros do not recurse (e.g., `#define sqrt(F) sqrt(fabs(F))’ is perfectly legal C).

Some other tidbits: 

The nyacc c99 module provides a few neat functions to support this.  One is expand-typerefs (was expand-decl-typerefs).  This will convert an sxml representation of 

  typedef const int (*foo_t)(int a, double b);
  extern foo_t fctns[2];

into one of 

  extern const int (*fctns[2])(int a, double b);

Another function, udecl->mspec, helps with the conversion as follows.  Consider

  double x[10]; /* state vector */

The c99 parser, followed by tree->udict, followed by cadr will turn the above into

  (udecl (decl-spec-list
          (type-spec (float-type "double")))
        (init-declr
          (array-of (ident "x") (p-expr (fixed "10"))))
        (comment " state vector “))

And “udecl->mspec/comm” will convert the above into

  ("x"
   " state vector "
   (array-of "10")
   (float-type "double”))

I am in the process of cleaning up the above in order to convert some function signatures.  

Matt





  parent reply	other threads:[~2017-03-14  0:53 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-03-09  2:06 ffi helper Matt Wette
2017-03-11 11:38 ` Taylan Ulrich Bayırlı/Kammer
2017-03-11 19:55   ` David Pirotte
2017-03-14  0:53 ` Matt Wette [this message]
2017-03-19 17:23   ` Matt Wette
2017-03-20 14:49     ` Taylan Ulrich Bayırlı/Kammer
2017-03-21 22:40     ` Matt Wette
2017-04-18  2:26       ` Matt Wette
2017-05-13 16:19         ` Matt Wette
2017-03-23 17:43 ` Amirouche
2017-03-24  1:43   ` Matt Wette

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

  List information: https://www.gnu.org/software/guile/

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=95B6EEEF-5278-44C8-B640-647A16AEFD1F@gmail.com \
    --to=matt.wette@gmail.com \
    --cc=guile-user@gnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).