Re: fix for time-date.el

Re: fix for time-date.el

[Reiner Steib]

On Tue, Dec 23 2008, Dave Love wrote:

> CL is needed for `declare'.  Also note the undefined function -- what is
> that?

Did you test with Emacs 21?  In Emacs 22 and up, declare is a Lisp
macro in `subr.el', assoc-string is a built-in function in
`src/minibuf.c'.

`format-seconds' (where `assoc-string' is used) is not used anywhere
in Gnus so it's not a real problem, AFAICS.  (time-date.el is use in
other parts of Emacs.)

> 2008-12-23  Dave Love  <fx@gnu.org>
>
> 	* time-date.el (cl): Require when compiling.
>
> Index: time-date.el
> ===================================================================
> RCS file: /usr/local/cvsroot/gnus/lisp/time-date.el,v
> retrieving revision 7.16
> diff -u -F^( -r7.16 time-date.el
> --- time-date.el	19 May 2008 08:47:42 -0000	7.16
> +++ time-date.el	23 Dec 2008 15:45:55 -0000
> @@ -39,6 +39,8 @@
>
>  ;;; Code:
>
> +(eval-when-compile (require 'cl))
> +
>  (defmacro with-decoded-time-value (varlist &rest body)
>    "Decode a time value and bind it according to VARLIST, then eval BODY.
>
> @@ -290,6 +292,7 @@ (defun format-seconds (string seconds)
>        (setq start (match-end 0)
>              spec (match-string 1 string))
>        (unless (string-equal spec "%")
> +	;; Fixme: assoc-string not defined.
>          (or (setq match (assoc-string spec units t))
>              (error "Bad format specifier: `%s'" spec))
>          (if (assoc-string spec usedunits t)

Bye, Reiner.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo---  |  PGP key available  |  http://rsteib.home.pages.de/

Re: fix for time-date.el

[Dave Love]

Reiner Steib <reinersteib+gmane@imap.cc> writes:

> Did you test with Emacs 21?

Yes.  I thought it was supposed still to be supported -- that's what the
gnus-coding Info file says.

> In Emacs 22 and up, declare is a Lisp
> macro in `subr.el', assoc-string is a built-in function in
> `src/minibuf.c'.

*Sigh*.  It's been in CL for a very long time...  There's a distressing
amount of that sort of thing in 22.

> `format-seconds' (where `assoc-string' is used) is not used anywhere
> in Gnus so it's not a real problem, AFAICS.

Fortunately the function where it's used seems to be new; otherwise it
would be a problem.  It seems to me to merit a comment at least.

Emacsen supported (was: fix for time-date.el)

[Dave Love]

I wrote:

> Reiner Steib <reinersteib+gmane@imap.cc> writes:
>
>> Did you test with Emacs 21?
>
> Yes.  I thought it was supposed still to be supported -- that's what the
> gnus-coding Info file says.

Could someone tell me whether it is still supposed to be supported to
avoid me wasting time on Emacs 21 support, like a different fix for this
inconsistent `declare' if requiring cl isn't acceptable?

Re: Emacsen supported

[Steinar Bang]

>>>>> Dave Love <d.love@liverpool.ac.uk>:

> I wrote:
>> Reiner Steib <reinersteib+gmane@imap.cc> writes:

>>> Did you test with Emacs 21?

>> Yes.  I thought it was supposed still to be supported -- that's what
>> the gnus-coding Info file says.

> Could someone tell me whether it is still supposed to be supported to
> avoid me wasting time on Emacs 21 support, like a different fix for
> this inconsistent `declare' if requiring cl isn't acceptable?

FWIW one of my Gnusen runs on a debian stable machine.  Stable is
currently etch, and emacs21 is the newest GNU emacs on it.

Re: fix for time-date.el

[Reiner Steib]

On Mon, Dec 29 2008, Dave Love wrote:

> Reiner Steib <reinersteib+gmane@imap.cc> writes:
>> Did you test with Emacs 21?
>
> Yes.  I thought it was supposed still to be supported -- that's what the
> gnus-coding Info file says.

Yes, it is supposed to be supported.  I don't test with Emacs 21 (nor
XEmacs) though.

BTW, the major reason not to drop support for Emacs 21 is that we
won't gain much because often the same code is necessary for XEmacs
21.4 (and 21.5).

>> In Emacs 22 and up, declare is a Lisp macro in `subr.el',
>> assoc-string is a built-in function in `src/minibuf.c'.
>
> *Sigh*.  It's been in CL for a very long time...  There's a distressing
> amount of that sort of thing in 22.

IIRC, several useful functions have been moved from CL to Emacs core.
(Maybe after this discussion:
<http://thread.gmane.org/gmane.emacs.devel/78898/focus=78981>)

>> `format-seconds' (where `assoc-string' is used) is not used anywhere
>> in Gnus so it's not a real problem, AFAICS.
>
> Fortunately the function where it's used seems to be new; otherwise it
> would be a problem.  It seems to me to merit a comment at least.

I've added a comment in `format-seconds' and (eval-when-compile
(require 'cl)) along with a comment.

Bye, Reiner.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo---  |  PGP key available  |  http://rsteib.home.pages.de/

Re: Emacsen supported

[Reiner Steib]

On Mon, Jan 05 2009, Dave Love wrote:

> I wrote:
>> Reiner Steib <reinersteib+gmane@imap.cc> writes:
>>> Did you test with Emacs 21?
>> Yes.  I thought it was supposed still to be supported -- that's what the
>> gnus-coding Info file says.
>
> Could someone tell me

I though I already responded, but the message was still in my drafts
folder.  Sorry.

> whether it is still supposed to be supported to avoid me wasting
> time on Emacs 21 support,

Yes, it is supposed to be supported.  I don't test with Emacs 21 (nor
XEmacs) though.  The major reason not to drop support for Emacs 21 is
that we won't gain much because often the same code is necessary for
XEmacs 21.4 (and 21.5).

> like a different fix for this inconsistent `declare' if requiring cl
> isn't acceptable?

I've committed this:

;; Only necessary for `declare' when compiling Gnus with Emacs 21.
(eval-when-compile (require 'cl))

Bye, Reiner.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo---  |  PGP key available  |  http://rsteib.home.pages.de/

Re: Emacsen supported

[Dave Love]

Reiner Steib <reinersteib+gmane@imap.cc> writes:

> Yes, it is supposed to be supported.  I don't test with Emacs 21 (nor
> XEmacs) though.  The major reason not to drop support for Emacs 21 is
> that we won't gain much because often the same code is necessary for
> XEmacs 21.4 (and 21.5).

I'm rather surprised.  Emacs 22 has sadly broken a lot of stuff, but I
doubt it's as far from Emacs 21 as XEmacs is.  The most important
difference is that XEmacs doesn't have the m17n facilities to cope with
MIME properly, or certainly didn't when I maintained Gnus, and not a lot
appears to have changed.

A better reason to support Emacs 21 seems to be that it's in currently
supported OSes, especially versions of GNU/Linux.