Dependency problem with auth-source and eieio

Dependency problem with auth-source and eieio

[Matt Lundin]

When trying to access the docstring of auth-source-backend with "C-h v
auth-source-backend", I get the following error:

--8<---------------cut here---------------start------------->8---
Debugger entered--Lisp error: (void-function cedet-called-interactively-p)
  cedet-called-interactively-p()
  eieio-describe-class(auth-source-backend)
  describe-variable(auth-source-backend)
  call-interactively(describe-variable nil nil)
--8<---------------cut here---------------end--------------->8---

It seems that eieio has a missing cedet-compat require here. I raise
this issue on the Gnus list, because eieio (without cedet) is now being
included in the Gnus source tree.

Note: I certainly hope we don't have to add a new dependency here, as it
would be a shame if we had to require cedet to compile/run Gnus. The
current eieio dependency/hack seems extreme enough as it is. :)

Best,
Matt

Re: Dependency problem with auth-source and eieio

[David Engster]

Matt Lundin writes:
> When trying to access the docstring of auth-source-backend with "C-h v
> auth-source-backend", I get the following error:
>
> Debugger entered--Lisp error: (void-function cedet-called-interactively-p)
>   cedet-called-interactively-p()
>   eieio-describe-class(auth-source-backend)
>   describe-variable(auth-source-backend)
>   call-interactively(describe-variable nil nil)
>
> It seems that eieio has a missing cedet-compat require here. I raise
> this issue on the Gnus list, because eieio (without cedet) is now being
> included in the Gnus source tree.
>
> Note: I certainly hope we don't have to add a new dependency here, as it
> would be a shame if we had to require cedet to compile/run Gnus. The
> current eieio dependency/hack seems extreme enough as it is. :)

cedet-called-interactively-p is used because Emacs versions <=23.1 don't
allow an argument for called-interactively-p; same goes for XEmacs. The
EIEIO from Emacs proper simply calls called-interactively-p, to which
cedet-called-interactively-p is an alias on Emacs versions >23.1.

Now, since EIEIO is only in Gnus because it should support those older
Emacs versions <=23.1, I'm afraid we have to provide that function. I
skimmed through cedet-compat.el, and I think it's at least currently the
only thing we need for EIEIO. The only other compatibility function that
EIEIO uses is find-coding-systems-region, which XEmacs doesn't
provide. This is only used though to save objects to disk, which we at
least currently do not do. But maybe it should be included, just to be
safe.

-David

Re: Dependency problem with auth-source and eieio

[David Engster]

Matt Lundin writes:
> When trying to access the docstring of auth-source-backend with "C-h v
> auth-source-backend", I get the following error:

I'm puzzled, though. You header says

> User-Agent: Gnus/5.110011 (No Gnus v0.11) Emacs/23.2 (gnu/linux)

Emacs 23.2 does have EIEIO, so I don't understand why you're seeing this
error. The EIEIO included in Emacs does not use
cedet-called-interactively-p, so it seems the one that ships with Gnus
shadows the one from Emacs proper. This shouldn't happen.

-David

Re: Dependency problem with auth-source and eieio

[Matt Lundin]

David Engster <deng@randomsample.de> writes:

> Matt Lundin writes:
>> When trying to access the docstring of auth-source-backend with "C-h v
>> auth-source-backend", I get the following error:
>
> I'm puzzled, though. You header says
>
>> User-Agent: Gnus/5.110011 (No Gnus v0.11) Emacs/23.2 (gnu/linux)
>
> Emacs 23.2 does have EIEIO, so I don't understand why you're seeing this
> error. The EIEIO included in Emacs does not use
> cedet-called-interactively-p, so it seems the one that ships with Gnus
> shadows the one from Emacs proper. This shouldn't happen.

Sorry, this was my mistake. The problem was caused by an older version
of cedet I had lying around in /usr/share/emacs/site-lisp. Once I
removed the offending cedet install, the problem went away. I can see
that the eieio included with emacs 23.2 and in the gnus git repo use
called-interactively-p instead of cedet-called-interactively-p.

I want to say thanks to Ted for all the work on the new auth-source. I'm
also very glad to have had an opportunity to explore eieio and cedet
further.

Best,
Matt

Re: Dependency problem with auth-source and eieio

[David Engster]

Matt Lundin writes:
> David Engster <deng@randomsample.de> writes:
>> Emacs 23.2 does have EIEIO, so I don't understand why you're seeing this
>> error. The EIEIO included in Emacs does not use
>> cedet-called-interactively-p, so it seems the one that ships with Gnus
>> shadows the one from Emacs proper. This shouldn't happen.
>
> Sorry, this was my mistake. The problem was caused by an older version
> of cedet I had lying around in /usr/share/emacs/site-lisp. Once I
> removed the offending cedet install, the problem went away. I can see
> that the eieio included with emacs 23.2 and in the gnus git repo use
> called-interactively-p instead of cedet-called-interactively-p.

Good; then the EIEIO version from Emacs was merged. But here the problem
is that certain features won't work with older Emacsen (writing objects
to disk, describe-variable on objects). Sorry that I wasn't thinking of
this earlier.

-David

Re: Dependency problem with auth-source and eieio

[Ted Zlatanov]

On Sun, 20 Feb 2011 11:48:44 -0500 Matt Lundin <mdl@imapmail.org> wrote: 

DE> Good; then the EIEIO version from Emacs was merged. But here the problem
DE> is that certain features won't work with older Emacsen (writing objects
DE> to disk, describe-variable on objects). Sorry that I wasn't thinking of
DE> this earlier.

Let's not worry about it now.  I'll complain if it's a problem later :)

Ted

Re: Dependency problem with auth-source and eieio

[Ted Zlatanov]

On Sun, 20 Feb 2011 11:48:44 -0500 Matt Lundin <mdl@imapmail.org> wrote: 

ML> I want to say thanks to Ted for all the work on the new auth-source. I'm
ML> also very glad to have had an opportunity to explore eieio and cedet
ML> further.

Thanks, and my apologies for the recent breakage.

Ted