Bug report ID:635627 - subst([...] is order-dependent

Am Donnerstag, den 01.04.2010, 15:11 -0600 schrieb Robert Dodier:
> On Thu, Apr 1, 2010 at 2:32 PM, Dieter Kaiser <drdieterkaiser at web.de> wrote:
> 
> > When I summarize the different opinions I think it is the best not to
> > change the current implementation of subst, to add some documentation
> > about the way substitute works, and to close the open bug report as
> > "won't fix".
> 
> OK.
> 
> > Adding a function like psubstitute for parallel substitution which has a
> > functionality comparable to the function substitute.
> 
> OK by me, but let's call it psubst to make it similar to the existing function.


A remark: The implementation is substitute. The name subst is defined as
an alias on the property list. Therefore, to do it similar we have to
implement psubstitute with an alias psubst.


> > Remark: I think it is not a good idea to try to extend sublis to get a
> > functionality like substitute, because we have to double a lot of code.
> 
> If we invent psubst to be like parallel version of subst,
> can we get rid of sublis ? (Or get sublis to punt to psubst?)
> I don't like functions which differ in subtle ways.


I will have a look at sublis to find the differences between an
implementation of psubst and sublis.


> Aside from the serial versus parallel difference which we are
> going to make explicit, there is also the question of how
> simplification occurs; that's really a mess. Sorry, I don't know
> how to clean it up.


I think it is not as complicated as it seems to be. Again I will have a
look in more detail at this topic.

Dieter Kaiser