Michael Olson <[EMAIL PROTECTED]> writes:

> I'm fairly certain that CL.el does not have any sort of union structure.

Ok.

> And even if it did, we would not want to use it if it would
> introduce a dependency on cl at run time (since RMS does not want
> code to (require 'cl)). It would probably be best just to leave that
> particular part of the structure unchanged.

I think it's too late for that. At least for the xmtn back-end; it
uses cl runtime extensively.

grepping dvc/lisp shows that tla-core and tla-defs also have "require
'cl" not inside "eval-when-compile". Although I'm not sure specifying
'eval-when-compile' really means there is no run-time dependence; I
think it more precisely means there _is_ a _compile-time_ dependence
(via macros).

RMS is stepping down as Emacs maintainer (sometime; I'm not reading
the emacs-devel list anymore, so I don't know if it has actually
happened yet). So maybe we could lobby for a change in this policy.
Since cl is distributed as part of Emacs, the policy doesn't make much
sense to me. It's like saying "we distribute diff-mode, but please
don't use it"!

If the argument is "cl is slow" then the response is "it's fast enough
for 90% of users" combined with "getting it right is more important
than making it really fast".

Allowing cl runtime does make the code easier to understand and write,
and therefore makes it easier to get right. Although it also makes it
more tedious to debug. I don't know how it affects executing speed; at
the moment there are more obvious things we could do to speed things
up (that's another several threads :).

If the policy does not change, we will need a distribution mechanism
that allows xmtn as an add-on. I certainly don't want to rewrite it to
get rid of the cl runtime.

Hmm. I suspect I'm also introducing cl runtime dependence in the DVC
front-end, by using ecase and defstruct in dvc-status.

If this is is a problem, I could go back to using cond instead of
ecase. I'm not clear if defstruct requires cl runtime; how do I tell?

--
-- Stephe


_______________________________________________
Dvc-dev mailing list
[email protected]
https://mail.gna.org/listinfo/dvc-dev

Reply via email to