Re: gnus shouldn't be making general-purpose variables buffer-local

[David Engster <deng@randomsample.de>]

Katsumi Yamaoka <yamaoka@jpl.org> writes:
>> On Thu, 11 Dec 2008 22:42:39 -0800 "Ami Fischman" <ami@fischman.org> wrote:
>>>>> Gnus makes 'timestamp a buffer-local variable in *Summary* buffers
>>>>> (and, I suspect, any other set group parameter).  This causes buggy
>>>>> interaction with non-gnus (async, usually) functions that bind
>>>>> 'timestamp as a function-local variable, such as emacs-jabber's
>>>>> jabber-history-log-message.  For the details please see
>>>>> http://thread.gmane.org/gmane.emacs.devel/105256.  Can gnus be changed
>>>>> to prefix such buffer-locals with at least gnus- or, better yet,
>>>>> gnus-summary-<group-name>- ?

There are indeed lots of possible group parameters which turn to
buffer-local variables. I am surely not innocent here, with nnmairix
having group parameters like `query', `folder' and `threads'. The
problem I see is that changing those is incompatible with existing
groups... not sure how to deal with that...

> Well, how can I get `timestamp' as a buffer-local variable in
> the summary buffer?  

I have some nnimap groups with `timestamp' in their group parameters,
but don't know which feature put it there. You can search for
`timestamp' in your .newsrc.eld.

> So far I found neither such a variable nor
> a way to make Gnus make it.
>
> C-h v timestamp RET
>  => [No match]

You would have to bind it globally first to see that it's buffer-local.

Regards,
David