bug#27436: 26.0.50; cc-defs requires cl instead of cl-lib in master

bug#27436: 26.0.50; cc-defs requires cl instead of cl-lib in master

[Alex]

cc-defs has this condition to check whether or not to include cl-lib:

======================================
(and (fboundp 'mapcan)
     (subrp (symbol-function 'mapcan)))
======================================

As of 466ee1b3ea, Emacs now has a built in mapcan, so the above returns
true.

Other code in cc-defs.el uses (featurep 'xemacs). Is there a reason why
it can't be used here?

bug#27436: 26.0.50; cc-defs requires cl instead of cl-lib in master

[Alan Mackenzie]

Hello, Alex.

On Wed, Jun 21, 2017 at 00:04:12 -0600, Alex wrote:
> cc-defs has this condition to check whether or not to include cl-lib:

> ======================================
> (and (fboundp 'mapcan)
>      (subrp (symbol-function 'mapcan)))
> ======================================

> As of 466ee1b3ea, Emacs now has a built in mapcan, so the above returns
> true.

Yes.  Thanks for taking the trouble to report this bug.

I've committed a fix to master.  I'd be grateful if you would confirm it
fixes the problem properly, or tell me what's not quite right.

> Other code in cc-defs.el uses (featurep 'xemacs). Is there a reason why
> it can't be used here?

CC Mode has always favoured testing features rather than version numbers
and the the base (X)Emacs, so that if a tested feature becomes or ceases
to be available, CC Mode should pick that up automatically.  That is
illustrated by the current case, just that here the actual test used was
suboptimal.  Hopefully it is now OK.

-- 
Alan Mackenzie (Nuremberg, Germany).

bug#27436: 26.0.50; cc-defs requires cl instead of cl-lib in master

[Alex]

Alan Mackenzie <acm@muc.de> writes:

> Hello, Alex.
>
> On Wed, Jun 21, 2017 at 00:04:12 -0600, Alex wrote:
>> cc-defs has this condition to check whether or not to include cl-lib:
>
>> ======================================
>> (and (fboundp 'mapcan)
>>      (subrp (symbol-function 'mapcan)))
>> ======================================
>
>> As of 466ee1b3ea, Emacs now has a built in mapcan, so the above returns
>> true.
>
> Yes.  Thanks for taking the trouble to report this bug.
>
> I've committed a fix to master.  I'd be grateful if you would confirm it
> fixes the problem properly, or tell me what's not quite right.
>
>> Other code in cc-defs.el uses (featurep 'xemacs). Is there a reason why
>> it can't be used here?
>
> CC Mode has always favoured testing features rather than version numbers
> and the the base (X)Emacs, so that if a tested feature becomes or ceases
> to be available, CC Mode should pick that up automatically.  That is
> illustrated by the current case, just that here the actual test used was
> suboptimal.  Hopefully it is now OK.

Hello, Alan.

Thanks for the explanation and fix; I can confirm that the problem is
fixed in master. I'll close this bug.