[Ecls-list] Megapatches ahead

Matthew Mondor mm_lists at pulsar-zone.net
Wed May 5 20:48:20 UTC 2010


On Wed, 5 May 2010 22:30:05 +0200
Juan Jose Garcia-Ripoll <juanjose.garciaripoll at googlemail.com> wrote:

> On Wed, May 5, 2010 at 10:18 PM, Matthew Mondor <mm_lists at pulsar-zone.net>wrote:
> 
> > On Wed, 5 May 2010 22:00:45 +0200
> > Juan Jose Garcia-Ripoll <juanjose.garciaripoll at googlemail.com> wrote:
> >
> > > Those functions are only used when the type of the arguments or of the
> > > output is not known or it does not match any of the C integer and
> > floating
> > > point types.
> > >
> > > For example in the code below ECL unboxes the value of X and does all
> > other
> > > operations in floating point and integer arithmetics.
> > [...]
> >
> > This is great to know, thanks :)
> >
> 
> But do not be too optimistic: right now the problem is that the compiler
> does not allow for simple type propagation due to the limited intermediate
> representation. That is one of the things I am trying to solve.

Before posting about inlining, I had existing code using a structure
with fixnum slots, and in the resulting code only generic functions were
used, but the code also lacked declarations before using the slots.  I
guess that the type progagation you're talking about will also affect
such cases?  Or would explicit declarations still be required before
using such slots (it was also the case with SBCL when I started using
it, but that apparently changed recently)?

Thanks again,
-- 
Matt




More information about the ecl-devel mailing list