Re: is requiring cl bad?

[David De La Harpe Golden]

On 17/12/12 19:09, Pascal J. Bourguignon wrote:
> > 24.3 finally provides an alternative: `cl-lib' which offers the
> > same functionality but in a namespace-clean way (i.e. using a "cl-"
> > prefix everywhere).
>
> This is a silly solution.
> The right solution is to implement a package system.

The consistent-prefix approach may be a C-like solution, but is still 
noticeably better than the previous conflicty mess, and widely
used for other emacs lisp libs.

But regarding the idea of "a package system" in particular, as you may 
mean a system similar to common lisp "packages":

If emacs ever did go toward adding new facilities in the general area of 
modularity (however unlikely it is in reality in the near future), I 
reckon Ron Garret's common lisp land "lexicons" work [1] might be a 
better "lispy modularity thingy" for emacs lisp to be inspired by than 
common lisp packages in particular.  At least, I'd take a hard look at 
lexicons (and at least glance at what some other languages do), before 
just blindly adding common lisp style packages.

Emacs lisp is lexically scoped now after all.

To people coming from quite a few other languages with more 
[nowadays-]conventional module systems, lexicons might well seem less 
weird than packages.

In emacs lisp land there is obviously no prior usage of common lisp 
style packages in the first place (yes I am vaguely aware you can feck 
about with non-default obarrays in emacs lisp for some purposes, but 
meh), unlike the situation in common lisp land where inertia and  compat 
concerns will probably keep most people on packages anyway despite the 
existence of lexicons there now.

[1] http://www.flownet.com/ron/lisp/lexicons.pdf
"""
Lexicons are first-class global lexical environments, what are sometimes 
called "modules" or "namespaces" in other languages.
[...]
instead of mapping strings to symbols, lexicons map symbols to (global) 
bindings.
"""