Progress migrating cffi and pycparser to libclang

Armin Rigo armin.rigo at gmail.com
Fri Jan 5 05:59:05 EST 2018


Hi Etienne,

On 5 January 2018 at 10:15, Etienne Robillard <tkadm30 at yandex.com> wrote:

> Forwarding  this thread to the CFFI developers...
>
If you're asking whether we could add libclang as a dependency to CFFI, the
answer is no, sorry.

I feel that I've already explained exactly this to you several times in
private e-mails, so I'm writing down this fact here on the public
python-cffi mailing list.  Please stop asking the same question.

For reference, here's my answer again.  CFFI is meant to be used in a
certain way.  I know from experience that a few people keep thinking about
it in a different way---sometimes for good reasons: it is tedious to wrap a
very large library even if it is easy to copy-paste-tweak individual
function definitions.  So to automate the process these people usually want
to parse real .h files, using libclang or better pycparser integration or
something else; and that's cool.  As I said there are already third-party
projects that try to do something similar.  Usually it is never completely
general, but more focused on a particular style of .h files coming from a
particular large project.  Maybe you'll run into similar issues and make
something that works in your case but not in others.  Or maybe you'll
manage to make it general enough.  In all cases this won't be included
inside CFFI, but could nevertheless be a successful separate project.
Please think about it as a project *above* CFFI, maybe something you'll
import and use inside a CFFI build script in order to provide content to
ffi.cdef().


A bientôt,

Armin.



More information about the Python-list mailing list