On Fri, 29 Jan 2010 13:59:27 -0600 Ted Zlatanov wrote: TZ> On Thu, 14 Jan 2010 10:38:44 -0600 Ted Zlatanov wrote: TZ> On Thu, 14 Jan 2010 10:44:13 -0500 Stefan Monnier wrote: >>>> Simon Josefsson already put a patch together at >>>> http://josefsson.org/securemacs but it will probably need to be revised >>>> a bit, the last change was in 2002. It offers gnutls.el as an >>>> alternative to starttls.el, with a similar API. As long as this is >>>> optional and autodetected through configure, I don't see a downside. It >>>> may need to be folded into starttls.el but that's not a big deal. I'm >>>> cc-ing Simon in case he has any comments. SM> I think I'd be OK with adding a --with-gnutls configure option that SM> links Emacs against that library. Especially if we can provide SM> a compatibility footls.el library that works both --with-gnutls SM> and without. TZ> Is anyone interested in reviewing Simon's patch? I think it does TZ> exactly what you suggest except the compatibility layer, which should be TZ> trivial since Simon kept the invocation semantics the same between TZ> gnutls.el and starttls.el. I posted a revised version of the patch on the gnutls-devel mailing list and asked for help there: http://thread.gmane.org/gmane.comp.encryption.gpg.gnutls.devel/4430 (note there's a minor revision of the patch posted today at http://thread.gmane.org/gmane.comp.encryption.gpg.gnutls.devel/4442) Simon Josefsson is active on that list but doesn't seem interested in further supporting that patch. So it's up to the Emacs developers to take this on. The patch is far from done but at least all the wrapper code is written and it has no issues AFAIK. All that remains is for someone with good C knowledge to look through process.c and process.h and adjust the API calls appropriately. Unfortunately I haven't done C in a long time and don't know the Emacs internals well, so it's really inefficient for me to dig through it. I'm attaching the latest patch here for completeness. Ted