Adrian Aichner <aichner(a)ecf.teradyne.com> writes:
> Hello auto-save experts!
>
> Having trouble with auto-save on Windows NT natively build XEmacs, I
> found some facts and more questions.
>
> ---------------------------------------------------------------
> auto-saving in XEmacs NT native build does not work by default:
> ---------------------------------------------------------------
>
> auto-save-directory
> "c:\\Winnt\\Profiles\\AichnerA\\…
[View More]autosave\\"
>
> will cause an illegal
>
> buffer-auto-save-file-name
> "c:\\Winnt\\Profiles\\AichnerA\\autosave\\#c:\\Winnt\\Profiles\\AichnerA\\News\\drafts\\drafts\\1#"
Is that the default? Why is auto-save-directory set to anything other
than nil? It defaults to nil under Unix.
Perhaps under Windows we should default `auto-save-hash-p' so that
setting `auto-save-directory' works? But I don't like doing that if
there is another solution, because the docs say that auto-save-hash-p
defeats `recover-all-files'.
> Workaround:
> Setting auto-save-directoy to nil, which makes files being auto-saved
> in the same directory as buffer-file-name, making file names legal
> when TEMP or TMP contain a drive spec.
That "workaround" is actually the default behaviour under Unix.
> Question:
> How could this problem be fixed? I suppose buffer-auto-save-file-name
> should have any embedded drive:\ or \\computer-name component
> stripped?
No, because then you wouldn't be able to differentiate between the
autosave for c:\foo and the autosave for d:\foo.
> (defcustom auto-save-directory nil
>
> ;; Don't make this user-variable-p, it should be set in .emacs and
> ;; left at that. In particular, it should remain constant across
> ;; several Emacs session to make recover-all-files work.
>
> ;; However, it's OK for it to be customizable, as most of the
> ;; customizable variables are set at the time `.emacs' is read.
> ;; -hniksic
>
> I fail to see how and why a user of the trusting kind should have to
> figure out that his customization of auto-save-directory (even
> though one can set it for the current session) does not get saved in
> (custom-set-variables ...)
What on earth do you mean by that? Are you sure you understood the
quoted comment?
> Only by looking at the above source-code will one get the hint that
> one has to (setq auto-save-directory nil) in ones .emacs file.
`M-x customize-variable RET auto-save-directory RET' should also work.
> Question:
> Why shouldn't the following initialization code for auto-save-interval
>
> (put 'auto-save-interval 'custom-type 'integer)
> (put 'auto-save-interval 'factory-value '(300))
> (custom-add-to-group 'auto-save 'auto-save-interval 'custom-variable)
>
> be made into a plain (defcustom ...)?
Is `defcustom' defined at the time auto-save is loaded?
> Question:
> Why can't auto-save-timeout be a (defcustom ...)
auto-save-timeout *is* a customizable variable. I'm not sure why it's
not initialized via `defcustom'. However, `M-x customize-variable RET
auto-save-timeout RET' will do the right thing.
--
Hrvoje Niksic <hniksic(a)srce.hr> | Student at FER Zagreb, Croatia
--------------------------------+--------------------------------
If anything goes wrong, blame the random-number generator!
[View Less]
> Thanks for nominating me, it was an honor. Unfortunately, none of us
> (me, Erik, Lars, Jamie) even made finalist. :-( boo hoo ...
>
> http://lwn.net/daily/fsa-finalists.html
There were a lot of good people on the original list. So I wouldn't
feel too bad about not making the finals. I think it will take
several years before all the people that deserve to win, actually
make it to the finals.
My guess is that either Donald Knuth or Larry Wall will win.
--
Richard Coleman
coleman(a)math.gatech.edu
>>>>> "APA" == Adrian Aichner <aichner(a)ecf.teradyne.com> writes:
>>>>> "SL" == SL Baur <steve(a)xemacs.org> writes:
SL> A patch against XEmacs 21.0-pre14 "Poitou" is in
SL> ftp://ftp.xemacs.org/pub/xemacs/beta/testing/
APA> Hi Steven,
APA> I downloaded xemacs_codename="Pyrenean-pre2" via CVS and
APA> found conflicts between my lisp\package-get-base.el[1] and
APA> the version in the repository.
APA…
[View More]> After sorting the repository version by use of
APA> (sort-regexp-fields nil "^(\\(.+\\)\\( [^)].*\\)+\\( )) \\)"
APA> "\\1" (region-beginning) (region-end))
APA> I ediff-ed it against my fairly up-to-date version, sorted
APA> the same way.
APA> The diffs are frightening:
APA> package sizes are off by very few bytes. Consequently all
APA> md5sums are different.
Steven,
I just checked a few packages in
/ftp@ftp.xemacs.org:/pub/xemacs/packages:
and all sizes indicate package-get-base.el in Pyrenean-pre2 to be
wrong.
I'll try building now with my package-get-base-latest.el of yesterday.
Regards,
Adrian
APA> Please comment.
APA> Adrian
APA> Footnotes: [1] Which did not come from pleasure but through
APA> pain. I have merged in package-get-base-update of few days
APA> ago and package-get-base.incremental-1998-10-01
--
Adrian Aichner
Applications Engineer
Teradyne GmbH
Semiconductor Test Group Telephone +49/89/41861(0)-208
Dingolfinger Strasse 2 Fax +49/89/41861-217
D-81673 MUENCHEN E-mail adrian.aichner(a)teradyne.com
[View Less]
[I'm working towards a goal of removing the GNU make dependency in
building the XEmacs Lisp tree from source, but it won't go very fast.]
There are now methods of installing executable scripts (but not
binaries) from lisp packages. Name them with $(LIBSRC_FILES).
The many binkit-* rules have been obsoleted by a single binkit-common
rule.
The Gnats and Tm scripts have been removed for the next XEmacs
releases. The `mmencode' binary remains because I don't have a good
silver bullet to kill …
[View More]it with.
As Hrvoje has pointed out, etags/ctags/ootags/etags.el needs to be
packaged too, but they will require very specialized support beyond
basic binary building. I haven't begun to deal with this issue, but
welcome suggestions.
Setting $(STAGING) to something appropriate and commenting out
$(MAKE_TARBALLS) should now work as a `make install' into a live
XEmacs package lisp tree.
I'm going to get these changes checked in by sometime late tomorrow
evening (US/Pacific) after I have completed testing them.
[View Less]
Hello auto-save experts!
Having trouble with auto-save on Windows NT natively build XEmacs, I
found some facts and more questions.
---------------------------------------------------------------
auto-saving in XEmacs NT native build does not work by default:
---------------------------------------------------------------
auto-save-directory
"c:\\Winnt\\Profiles\\AichnerA\\autosave\\"
will cause an illegal
buffer-auto-save-file-name
"c:\\Winnt\\Profiles\\AichnerA\\autosave\\#c:\\Winnt\\…
[View More]Profiles\\AichnerA\\News\\drafts\\drafts\\1#"
Workaround:
Setting auto-save-directoy to nil, which makes files being auto-saved
in the same directory as buffer-file-name, making file names legal
when TEMP or TMP contain a drive spec.
Question:
How could this problem be fixed? I suppose buffer-auto-save-file-name
should have any embedded drive:\ or \\computer-name component stripped?
Question:
Although I found this in auto-save.el
(defcustom auto-save-directory nil
;; Don't make this user-variable-p, it should be set in .emacs and
;; left at that. In particular, it should remain constant across
;; several Emacs session to make recover-all-files work.
;; However, it's OK for it to be customizable, as most of the
;; customizable variables are set at the time `.emacs' is read.
;; -hniksic
I fail to see how and why a user of the trusting kind should have to
figure out that his customization of auto-save-directory (even though
one can set it for the current session) does not get saved in
(custom-set-variables ...)
Only by looking at the above source-code will one get the hint that
one has to (setq auto-save-directory nil) in ones .emacs file.
The doc-string could still warn the user about changing this variable
thoughtlessly, even if it is made into a (defcustom ...).
Question:
Why shouldn't the following initialization code for auto-save-interval
(put 'auto-save-interval 'custom-type 'integer)
(put 'auto-save-interval 'factory-value '(300))
(custom-add-to-group 'auto-save 'auto-save-interval 'custom-variable)
be made into a plain (defcustom ...)?
Question:
Why can't auto-save-timeout be a (defcustom ...) so that it can
easily be Set and Saved?
Here is what I found in itimer-autosave.el:
;jwz: this is preloaded so don't ;;;###autoload
(defvar auto-save-timeout 960
Thanks for any insight.
I'm willing to send patches in case there is agreement about any
changes that should be made.
Best regards,
Adrian
--
Adrian Aichner
Applications Engineer
Teradyne GmbH
Semiconductor Test Group Telephone +49/89/41861(0)-208
Dingolfinger Strasse 2 Fax +49/89/41861-217
D-81673 MUENCHEN E-mail adrian.aichner(a)teradyne.com
[View Less]
A patch against XEmacs 21.0-pre14 "Poitou" is in
ftp://ftp.xemacs.org/pub/xemacs/beta/testing/
If anyone can tell me how to make `cvs rdiff' work against a branch,
I'll make a patch for 21.2 as well.
Highlights:
to 21.2 beta3 "Aglaia"
-- case sensitiveness improvements from Didier Verna
-- Bug fixes from 21.0
-- Word selection on mouse click on quotes from Hrvoje Niksic
-- WAVE support for NAS from Raymond Toy
to 21.0 "Pyrenean"
-- BS/DEL fix from Glynn Clements
-- Tty vertical divider …
[View More]width now defaults to 1 instead of 0
-- MS Windows toolbar fixes from Jonathan Harris
-- Fix for XtDisplay being called with stale widget from Takeshi Hagiwara
-- Fixes for Cygwin b20 from Andy Piper
-- Package UI fetching improvements from Greg Klanderman
-- isearch regexp pruning bugfix from Hrvoje Niksic
-- package deletion fixes from Jan Vroonhof
-- gnuclient fix from Jan Vroonhof
[View Less]
On my system, at least, there is a problem with a Windows-related
patch to gnuclient.c (by Raymond Toy) which was included in the
patch from 21.0-b56 to 21.0-b57. Executing 'gnuclient -q
[filename]' would bring up a new XEmacs window, but rather than
the file requested appearing in the window there would be a dired
listing of the directory. I tried recompiling XEmacs-21.0-b57
with the b56 version of gnuclient.c substituted for the patched
version, and gnuclient works as it should. This is on …
[View More]a Linux
2.1.125 glibc system.
Larry Ayers
[View Less]
I've uploaded a test XEmacs package of Mailcrypt-3.5b7 which has
support for GNU Privacy Guard (the GNU PGP replacement) and PGP 5.
The package name was changed to mailcryptng because I'm not ready to
replace mailcrypt-3.4 yet and on first glance some XEmacs support in
mailcrypt-3.4 was removed.
(mailcryptng
(standards-version 1.0
version "1.0"
author-version "3.5b7"
date "1998-10-10"
build-date "1998-10-10"
maintainer "XEmacs Development Team <xemacs-beta(a)xemacs.org…
[View More]>"
distribution stable
priority low
category "comm"
dump nil
description "Support for messaging encryption with PGP."
filename "mailcryptng-1.0-pkg.tar.gz"
md5sum "12c507b0e27fe032be9ad254f4c5c765"
size 120084
provides (mailcrypt)
requires (gnus vm mail-lib fsf-compat xemacs-base)
type regular
))
[View Less]