We are slowly heading towards a PSYC specification release, now that secure share is entering prototypical status. We'll have to decide whether calling it 1.0 is appropriate or an insult for a technology that has evolved over twenty years and is reaching the intended status of perfection only now.

Yes, PSYC always aimed at perfection, that's why it took twenty years. It's a bad idea to strive for perfection, but now that we're about to have it, we should rejoice. Most popular technologies are far from being perfect in what they do, so something that tries its best at it is good for a change.

Until now, PSYC did some things better and some things worse than other technologies. A psyced installation does a good job in production use and everyday life, and has been doing so for over a decade now. Still, LPC led us to a technological dead end, so we had a serious strategy discussion in 2009 that brought us bright new code written in compileable languages. libpsyc, psycd and Saikound.


Things left to be done in general:

  • Finish the prototype for secure share.
  • Fix websites so that they do a good job and can resist a slashdot or other sudden attention spike.

Things to be done in the specification:

  • Entity orientation in psyctext so all the stupid _nick* vars can be left out of the protocol (currently in #define EXPERIMENTAL).
  • The big keyword naming revamp, implement compact mode and become fully inheritance compliant (all issues depending on each other, so need to be solved together).

What's up with psyced:

After the big release:

  • We can specify all nice optional extensions one by one.
  • Let's publish a stable interserver PSYC protocol first. Later we provide a stable client protocol interface, which is still optional for PSYC to already serve a purpose in a world full of IRC and XMPP applications in need of a better interserver infrastructure. It is certainly worth separating, as native clients require a lot more than interserver.


Things done since I made this page.

See also