Re: sequence manipulation functions

[Stefan Monnier <monnier@iro.umontreal.ca>]

> I understand your point.  Then what do you think about having all other
> sequence functions prefixed with "seq-"?

Yes, you'll want to add seq-* aliases for things like elt.

> For backward-compatibility, I could keep the current names as aliases,
> and maybe later obsolete them.

We can start by having seq-elt be an alias for elt (rather than the
other way around) so as not to need changing anything else.

>> The different of cost between "rest of a list" and "rest of an array" is
>> so large that merging the two this way is probably not a good idea.
> Do you mean that the implementation is lacking in some way or that you
> would remove the "rest" function all together?

I think it's not just an implementation detail but a fundamental
limitation, so I'd just remove it.  Those who really want it can use
subseq (or whatever its new name will be, e.g. seq-subseq).

Sebastien Vauban <sva-news@mygooglest.com> writes:
> Why?  Isn't it still useful for packages in general (and ELPA in
> particular) so that autoloads files are generated, and these only are
> what's needed to be loaded... allowing performance gains at startup?

You might as well put a

   ;;;###autoload (require 'seq)

at that point (or add it to loadup.el).


        Stefan