Re: [Sbcl-devel] More host's gensyms

[Charles Z. <cha...@ya...>]

I came to a similar conclusion as Stas; the problem is that the host macroexpander could cause either the host gensym counter or the xc gensym counter to get incremented a different amount of times depending on the implementation, and so you leak that no matter which version of gensym you decide to unconditionally use, hence why they actually have to be differentiated.

 
 
  On Fri, Jan 22, 2021 at 15:36, Stas Boukarev<sta...@gm...> wrote:   On Fri, 22 Jan 2021 at 22:53 Manuel Giraud <ma...@le...> wrote:

Stas Boukarev <sta...@gm...> writes:

> Some time ago I tried shadowing gensym, but it didn’t completely help, as
> macros are allowed to be expanded multiple times. The only thing that
> worked was to always use zero. But that obviously is a bad idea.

(sure you did not want this on list?)

Yeah.

And do you think that chasing down leaky gensyms as I am doing could
help in place of shadowing?



I haven’t actually followed closely the discussion, just an observation._______________________________________________
Sbcl-devel mailing list
Sbc...@li...
https://lists.sourceforge.net/lists/listinfo/sbcl-devel