Re: rfc2047-decode-string in C?

[Stefan Monnier]

>> Actually, I think Emacs could do with moving more things down into
>> the C layer.

> Speaking of which -- I've been elp-ing Gnus group entry a bit, and Gnus
> spends quite a but of time doing rfc2047 decoding.  If I read the
> results correctly, that's a quite significant portion of the time.

> rfc2047-encoding looks like this:

> =?iso-8859-1?q?fo=F2o?=

> There are two encodings -- b and q.  The b (base64) is already in C
> (was I the one that put it there?  Hm.), but the q isn't
> (quoted-printable), so I'd propose two new C functions:
> `quoted-printable-decode-string' and `rfc2047-decode-string'.

> I can implement them and do some benchmarking?

A version of quoted-printable-decode-string written in C would be
fine, yes.  For rfc2047-decode-string, I'd need some convincing evidence
that there is a large benefit.


        Stefan "who uses Gnus and sometimes finds it quite slow, tho the main
                slowness is when I do gnus-group-get-new-news over IMAP"