[Ecls-list] Fwd: Slow prettyprinting
Juan Jose Garcia-Ripoll
jjgarcia at users.sourceforge.net
Mon Jan 14 21:25:38 UTC 2008
On Jan 14, 2008 9:04 PM, Waldek Hebisch <hebisch at math.uni.wroc.pl> wrote:
> So, change to caching has measurable positive effect on prettyprinting.
> But the effect is rather small. Given amount of consing I suspect
> that the bottlenect is somewhere else. Also, sbcl conses about
> 100 time less then ecl.
Waldek, I did not mean that caching would dramatically improve it, but
it is a first step. Consing comes probably from 1) the creation of
intermediate streams and structures for creating the pretty output and
2) lisp code calling the library and being called by it. The second
step is more subtle and related to why function calls to interpreted
functions are more expensive. In any case, I am very thankful for the
code you provide. It will help me tracking these issues.
Facultad de Fisicas, Universidad Complutense,
Ciudad Universitaria s/n Madrid 28040 (Spain)
More information about the ecl-devel