clisp-devel

subrkw.d & init_subr_tab_2

[Sam S. <sd...@gn...>]

Hi Bruno,

It appears that the only reason we have subrkw.d in addition to subr.d is:

 --- TURBO-C doesn't get enough memory for compilation!

(comment in spvw.d:init_subr_tab_2).
I tried to enable that code and - lo and behold - it worked!
(Well, after I fixed some bitrot :-).
So, we can get rid of subrkw.d now!
The new code generates a larger run time (by a 0.2%, or 4k)
and a larger memory image (by 0.1%, or 3.5k) though.
This is because subrkw allows subrs to share keyword vectors, thus
eliminating the code to create identical vectors and memory to hold
them.


-- 
Sam Steingold (http://sds.podval.org/) on CentOS release 5.6 (Final) X 11.0.60900031
http://jihadwatch.org http://camera.org http://openvotingconsortium.org
http://palestinefacts.org http://memri.org http://truepeace.org
Only a fool has no doubts.

Re: subrkw.d & init_subr_tab_2

[Bruno H. <br...@cl...>]

Hi Sam,

> It appears that the only reason we have subrkw.d in addition to subr.d is:
> 
>  --- TURBO-C doesn't get enough memory for compilation!

Yes, that was the main reason.

> So, we can get rid of subrkw.d now!
> The new code generates a larger run time (by a 0.2%, or 4k)
> and a larger memory image (by 0.1%, or 3.5k) though.

This optimization is not worth keeping. Earlier, without CLOS and
tons of other stuff in the memory image, the savings were higher.
But now...

Bruno
-- 
In memoriam Anne Boleyn <http://en.wikipedia.org/wiki/Anne_Boleyn>