Re: [PATCH] Two issues with the gnus-registry

[Eric Abrahamsen <eric@ericabrahamsen.net>]

Katsumi Yamaoka <yamaoka@jpl.org> writes:

> On Thu, 18 Dec 2014 23:09:41 +0800, Eric Abrahamsen wrote:
>> Looks like something already did! I don't know why though. Compiling
>> gnus on its own works fine, but emacs compilation does indeed break with
>> the "Invalid slot type: max-size integer most-positive-fixnum".
>
> This should have been fixed, thanks to Paul Eggert.

That's a relief it wasn't more complicated. I'm still not clear why I
was able to compile the file from within Emacs with no errors.

> But the Gnus buildbot[1] still complains there are glitches for
> compiling with old Emacsen (including XEmacsen).  Maybe it is
> due to some new cl macros.  Here is a compilation log I got in
> my 32-bit Cygwin environment:
>
> In registry-db::registry-collect-prune-candidates:
> registry.el:375:31:Warning: reference to free variable `for'
> registry.el:375:35:Warning: reference to free variable `k'
> registry.el:375:37:Warning: reference to free variable `being'
> registry.el:375:43:Warning: reference to free variable `the'
> registry.el:375:47:Warning: reference to free variable `hash-keys'
> registry.el:375:57:Warning: reference to free variable `of'
> registry.el:376:31:Warning: reference to free variable `using'
> registry.el:376:50:Warning: reference to free variable `v'
> registry.el:377:31:Warning: reference to free variable `when'
> registry.el:378:31:Warning: reference to free variable `collect'

This section of the code was always odd. I couldn't edebug it, either:
it gave me "bad `using' clause." It might be safer just to use a plain
`maphash'.

> In end of data:
> registry.el:387:1:Warning: the following functions are not known to be
>     defined: cl-remf, cl-loop, hash-values, cl-subseq
> Wrote /Work/gnus/lisp/registry.elc

These parts of the code weren't changed though -- surely we were getting
the same complaints all along?

Eric