Yes, that approach seems sane, though it would be interesting to see what others think about it.
I will be sending you a link to my github irssi repo, because (and sorry for that) I've been more busy than expected this week. It should contain the commits so you can at least start comparing my code to yours. I'm not a C genius, and I love constructive feedback, so don't hesitate to point out what sucks and what's good about my version of the patch. On Tue, Nov 11, 2014 at 2:04 PM, <[email protected]> wrote: > On Mon, Nov 10, 2014 at 02:37:19PM -0500, Guillaume Parent wrote: > > Hi, > > Yes, my client is a patch over 0.8.16-rc1 if I recall correctly. I am > > busy tonight but can possibly share the patch when I'm home. > > > I think it's an annoying balance between having a generic name for the > > option if the irssi dev team wants to use a different hash in the > > future vs. the clarity of the existing option. I had an attempt where > I > > used the length of the fingerprint even though that's pretty > > inaccurate. It worked during my testing but I didnt evolve it beyond > > that. Maybe something like how you specify password hashes? > > Good idea. Maybe something like > > ssl_fingerprint="sha256:DE:AD:BE:EF:..." > > Irssi passes first token to EVP_get_digestbyname so user can switch to > new type of hash as soon as it is available in Libre^WOpenSSL. > >
