Re: problem with gnus-group-make-nnir-group

["Jose A. Ortega Ruiz" <jao@gnu.org>]

Hi Ted,

On Mon, Oct 31 2011, Ted Zlatanov wrote:


[...]

> jao> But both `gnus-registry-enabled' *and* `gnus-bound-and-true-p'
> jao> are undefined when nnir is loaded.  gnus-registry.el does define
> jao> the former, but not the latter function, and in my emacs session
> jao> (fboundp 'gnus-bound-and-true-p) returns nil.
>
> I don't know how that's possible.  nnir.el has
>
> (defvar gnus-registry-enabled)

The variable is declared, but still unbound: trying to use it will raise
an error (because it has no value).

> and loads gnus-util.el, which defines `gnus-bound-and-true-p'.

This is a macro defined inside an eval-when-compile: the packaged Emacs
comes with a compiled version (gnus-util.elc), and it's the compiled
file that gets loaded when (require 'gnus-util) is called.  The idiom of
wrapping macros in an eval-when-compile form is intended for macros that
are going to be used in the same file that defines them.

> So this problem should not happen, and does not happen on the Gnus
> Buildbot either.  Can you think of something that could be causing the
> problem for you but not for me and the Buildbot?

Unless i'm missing something, the only explanation i have is that you're
loading source elisp files instead of compiled byte code.  Could that be
the case?

Cheers,
jao
-- 
"People sometimes ask me if it is a sin in the Church of Emacs to use vi.
 Using a free version of vi is not a sin; it's a penance".
  -Richard Stallman