Re: (eval-after-load "info" ..) triggers on loading anything-info

[Reiner Steib]

On Sat, May 27 2006, Eli Zaretskii wrote:

>> From: Martin Schoenmakers <address@hidden>
>> Date: Thu, 25 May 2006 20:23:48 +0200
>> 
>> As of very recent CVS builds emacs mutters on startup about 'Recursive
>> `require' for feature `sb-info''. Some digging around revealed that
>> this in turn was triggered from loading emms-info.el due to an
>> (eval-after-load "info" '(require 'sb-info)). The recursive requires
>> are triggered by the same line, which lives just before the relevant
>> provide in sb-info.el.
>
> There are no files that go by these names (emms-info.el and
> sb-info.el) in the Emacs CVS.  So I think you need to talk to whoever
> wrote those two files.

No, the recent changes to `eval-after-load' lead to such kind of
problems whenever you write...

  (eval-after-load "bar" '(add-to-list 'bar-variable 'baz))

... and some (other) file provides `foo-bar'.  `add-to-list' is evaled
after `foo-bar' is provided; _not_ after `bar' is provided.  Thus
you'll get "variable is void: bar-variable".

See the recent reports by Katsumi Yamaoka (`w3m' vs `mime-w3m'[1]),
Romain Francoise (`ibuffer' vs. `this-is-not-ibuffer.el') [2] and me
(`tex' vs. `rs-tex') [3].

Bye, Reiner.

Footnotes: 
[1]  http://lists.gnu.org/archive/html/emacs-pretest-bug/2006-05/msg00304.html
     http://mid.gmane.org/b4mhd3delpi.fsf%40jpl.org
     (on emacs-pretest-bug)

[2]  http://mid.gmane.org/873beywfew.fsf%40pacem.orebokech.com
     (on emacs-devel)

[3]  http://mid.gmane.org/v9lksodaft.fsf%40marauder.physik.uni-ulm.de
     (on emacs-devel)
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo---  |  PGP key available  |  http://rsteib.home.pages.de/