12:58 a.m.
Steve Youngs <sryoungs(a)bigpond.net.au> writes:
This patch means that Gnus no longer requires fsf-compat.
Have you actually checked that the Gnus code no longer uses things
from fsf-compat? Last time I looked, Gnus (and other packages
developed independently or carelessly synched) introduced many subtle
FSF-isms that have gone in unnoticed because fsf-compat was silently
blessed by its inclusion in the SUMOs.
This change in particular is suspicious:
Index: lisp/gpg.el
[...]
-(require 'timer)
+(if (featurep 'xemacs)
+ (require 'itimer)
+ (require 'timer))
(eval-when-compile (require 'cl))
Unless `gpg.el' has changed significantly from the version I have,
this change cannot be correct. gpg.el uses timer functions all over
the place without checking whether they're defined.
You have to either *really* port the file over to XEmacs's API's or
leave the `require', thereby acknowledging the dependence on
`timer.el' and consequently fsf-compat. That's how it works.
1:10 a.m.
New subject: [AC] [PACKAGES (gnus)] Remove Gnus' dependence on fsf-compat + Jerry's macro patch
Hrvoje Niksic <hniksic(a)xemacs.org> writes:
Steve Youngs <sryoungs(a)bigpond.net.au> writes:
> This patch means that Gnus no longer requires fsf-compat.
Have you actually checked that the Gnus code no longer uses things
from fsf-compat? Last time I looked, Gnus (and other packages
developed independently or carelessly synched) introduced many subtle
FSF-isms that have gone in unnoticed because fsf-compat was silently
blessed by its inclusion in the SUMOs.
This change in particular is suspicious:
> Index: lisp/gpg.el
[...]
> -(require 'timer)
> +(if (featurep 'xemacs)
> + (require 'itimer)
> + (require 'timer))
> (eval-when-compile (require 'cl))
Unless `gpg.el' has changed significantly from the version I have,
this change cannot be correct. gpg.el uses timer functions all over
the place without checking whether they're defined.
FWIW, gpg.el is not part of Gnus, it is only a contrib/, and pgg now
supersede it. But fixing any FSFism in Gnus would useful, there might
be someone who compiles Gnus herself instead of using the XEmacs
package.
2:33 a.m.
New subject: [AC] [PACKAGES (gnus)] Remove Gnus' dependence on fsf-compat + Jerry's macro patch
|--==> "HN" == Hrvoje Niksic <hniksic(a)xemacs.org> writes:
HN> Steve Youngs <sryoungs(a)bigpond.net.au> writes:
>This patch means that Gnus no longer requires fsf-compat.
HN> Have you actually checked that the Gnus code no longer uses things
HN> from fsf-compat?
Yes, I did check. I'm only human, but hopefully I didn't miss too
many things.
HN> This change in particular is suspicious:
>Index: lisp/gpg.el
HN> [...]
>-(require 'timer)
>+(if (featurep 'xemacs)
>+ (require 'itimer)
>+ (require 'timer))
>(eval-when-compile (require 'cl))
HN> Unless `gpg.el' has changed significantly from the version I have,
HN> this change cannot be correct. gpg.el uses timer functions all over
HN> the place without checking whether they're defined.
I have added another file to the gnus package `dgnushack-xemacs.el'
which is similar to the upstream `dgnushack.el'. In that, I have...
(defalias 'timer-create 'make-itimer)
(defalias 'cancel-timer 'delete-itimer)
(defalias 'timer-activate 'activate-itimer)
(defalias 'timer-set-time 'set-itimer-value)
(defalias 'timer-set-function 'set-itimer-function)
(unless (fboundp 'timer-relative-time)
(defun timer-relative-time (time secs &optional usecs)
"Advance TIME by SECS seconds and optionally USECS microseconds.
SECS may be a fraction."
(let ((high (car time))
(low (if (consp (cdr time)) (nth 1 time) (cdr time)))
(micro (if (numberp (car-safe (cdr-safe (cdr time))))
(nth 2 time)
0)))
;; Add
(if usecs (setq micro (+ micro usecs)))
(if (floatp secs)
(setq micro (+ micro (floor (* 1000000 (- secs (floor secs)))))))
(setq low (+ low (floor secs)))
;; Normalize
(setq low (+ low (/ micro 1000000)))
(setq micro (mod micro 1000000))
(setq high (+ high (/ low 65536)))
(setq low (logand low 65535))
(list high low (and (/= micro 0) micro)))))))
`dgnushack-xemacs.el' gets loaded via the package's Makefile
$PRELOADS. I've got to admit that I'm not sure how comfortable I am
with this type of approach. To me, the whole gnus "lpath/dgnushack"
thing stinks of hiding problems rather than fixing them.
Norbert, take your finger off that release trigger for a bit, I need
to have a think about this.
--
|---<Steve Youngs>---------------<GnuPG KeyID: A94B3003>---|
| XEmacs - The only _______ you'll ever need. |
| Fill in the blank, yes, it's THAT good! |
|------------------------------<sryoungs(a)bigpond.net.au>---|
7:31 a.m.
New subject: [AC] [PACKAGES (gnus)] Remove Gnus' dependence on fsf-compat + Jerry's macro patch
Steve Youngs <sryoungs(a)bigpond.net.au> writes:
I have added another file to the gnus package
`dgnushack-xemacs.el'
which is similar to the upstream `dgnushack.el'. In that, I have...
(defalias 'timer-create 'make-itimer)
(defalias 'cancel-timer 'delete-itimer)
(defalias 'timer-activate 'activate-itimer)
(defalias 'timer-set-time 'set-itimer-value)
(defalias 'timer-set-function 'set-itimer-function)
(unless (fboundp 'timer-relative-time)
(defun timer-relative-time (time secs &optional usecs)
[...]
If you are going to use things like this, please use a "gnus-timer-"
prefix, i.e. put those functions in a separate namespace and fix the
file to use those replacement functions. Otherwise someone will load
Gnus and think that it somehow provides a proper replacement for FSF
timer functionality.
This is why `timer.el' (and others in fsf-compat) have been written to
begin with. The idea is, if we import FSF's function names, we should
either do it right or not at all.
7660
days inactive
7660
days old
3 comments
3 participants
participants (3)
-
Hrvoje Niksic -
Simon Josefsson -
Steve Youngs