[Ecls-list] Invalid C output at safety 0

[Juan Jose Garcia-Ripoll]

On Dec 19, 2007 8:48 PM, Waldek Hebisch <hebisch at math.uni.wroc.pl> wrote:
> (as of svn revision 160) should build using ECL at safety 0.
> However, I must say that performance (speed) is rather poor.

Safety 0 currently does not mean top speed. I had to eliminate a lot
of "optimizations" to make ECL more stable, such as full unboxing of
functions and some inline forms. It can also be due to an "abuse" of
CLOS. ECL's implementation is not particularly efficient at method
dispatch. So, yes, one still has to improve a lot on performance and
it is big software like this one that may help.

Juanjo

-- 
Facultad de Fisicas, Universidad Complutense,
Ciudad Universitaria s/n Madrid 28040 (Spain)
http://juanjose.garciaripoll.googlepages.com