Re: About CEDET, Completion, and compilers

[joaotavora@gmail.com (João Távora)]

Andreas Röhler <andreas.roehler@online.de> writes:

> Given that's true - what about dropping EIEIO and re-building
> everything in plain Emacs Lisp?

FWIW, Eric Schulte's web-server [1] uses EIEIO and has recently been
accepted in Emacs ELPA. Eric reports it to be an roughly an order of
magnitude faster than el-node [2], and it apparently creates an EIEIO
object for every request. The code looks clean enough, but I didn't see
any `defmethod's or complicated inheritance.

But I've also noticed that some dismiss EIEIO as some kind of overkill,
a cannon to kill a fly. That's sometimes true, though I frequently wish
later that I had started out with a cannon. That's why in CL I almost
always start with CLOS right off the bat.

So, I've read some of the EIEIO docs, and know it's not as powerful as
CLOS, but would like to hear some experience about when to use it,
actual downsides experienced, etc...

Using more EIEIO could possibly make some people (including me)
understand, approach, and accept CEDET better.

João

[1] https://github.com/eschulte/emacs-web-server
[2] http://eschulte.github.io/emacs-web-server/benchmark/