5:51 p.m.
A couple of problems with the cygwin version,
xemacs -vanilla,
- M-x customize-face default
Right-click on [State]. Select "Show all display specs"
-> Invalid function: "Show all display specs"
I would guess that Jan's customize fixes are responsible for this. Stephen?
- M-x toggle-buffer-file-coding-system
Doesn't seem to change buffer-file-coding-system
Previously used to toggle how CR-LF was handled.
Wait, I just confirmed that this is not working under 21.4.5 as
well.
I sometimes see the problem where it sometimes works and sometimes doesn't.
However,
a simple test works for me. It may be that having non-binary mounts affects
things.
Hmm. This could be a problem with my setup...
xemacs,
- customization of Font Family of default face is not honored. I had
previously customized it to "Letter Gothic MT". Under 21.4.5, it is
displayed as "Letter Gothic MT". Under 21.4.6-rc1, it is displayed as
"Courier New".
Ditto Jan's changes. I don't think this would be caused by my customize fix.
Stephen?
andy
5:02 p.m.
New subject: XEmacs 21.4.6-rc1 binary kit available
"Andy Piper" <andy(a)xemacs.org> writes:
> - M-x customize-face default
> Right-click on [State]. Select "Show all display specs"
> -> Invalid function: "Show all display specs"
I would guess that Jan's customize fixes are responsible for this. Stephen?
Seems unlikely to me.
> xemacs,
> - customization of Font Family of default face is not honored. I had
> previously customized it to "Letter Gothic MT". Under 21.4.5, it is
> displayed as "Letter Gothic MT". Under 21.4.6-rc1, it is displayed as
> "Courier New".
I don't have a windows build to reproduce this, however I can
reproduce the problem with 'gtk build.
In that case the problem is that customize is using font.el to do the
conversion and that does not recognize 'gtk as a device type and
silently fails.
Applying this fix makes it work
--- /local/xemacs/sunsite/xemacs-21.4/gtk/lisp/font.el~ Sun Nov 25 17:27:33 2001
+++ /local/xemacs/sunsite/xemacs-21.4/gtk/lisp/font.el Sun Nov 25 17:27:33 2001
＠＠ -99,6 +99,7 ＠＠
(defconst font-window-system-mappings
'((x . (x-font-create-name x-font-create-object))
+ (gtk . (x-font-create-name x-font-create-object))
(ns . (ns-font-create-name ns-font-create-object))
(mswindows . (mswindows-font-create-name mswindows-font-create-object))
(pm . (x-font-create-name x-font-create-object)) ; Change? FIXME
Could something similar be happening for Windows?
If so, how did customize work before this?
The cleanest way to solve this is to have make-face-size and
make-face-family use a device type specific routine a long the lines
of make-face-italic. That way the use of font.el can be dropped.
I don't think "my patch" is really causing
When XEmacs 21.4.6rc1 is built on my system --with-gtk then default
faces are setup correctly to GTK values, but when they are changed with a
subsequent M-x customize-face or custom-set-faces form the change has no
visible effect.
either!
If the set-face-background forms within gtk-init-device-faces
aren't
passed the gtk device as the optional locale argument (see patch) then
everything is hunkydory and subsequent customization of the default
faces *do* have a visible effect.
There is something else weird going on with the face frobbing. I am
going to investigate.
[time passes]
I think the problem is entirely similar to the problem we had before
with 'italic'.
Customize might need to call frob-face-property as well.
Until then,
+ ;; Removed device locales - a naive attempt to keep Custom happy.
This seems like a good idea for the moment. Gtk doesn't support
multiple device anyway AFAIK. However,
(set-face-foreground 'highlight
- (nth prelight (plist-get style 'text))
- device)
+ (nth prelight (plist-get style 'text)))
+ ;device)
This not quite 100% right. It really should have been adding a '(gtk
default) tag set from the beginning, i.e. make it
(set-face-foreground 'highlight
(nth prelight (plist-get style 'text))
nil '(gtk default))
etc
Jan
5:19 p.m.
New subject: XEmacs 21.4.6-rc1 binary kit available
"Andy Piper" <andy(a)xemacs.org> writes:
> > - M-x customize-face default
> > Right-click on [State]. Select "Show all display specs"
> > -> Invalid function: "Show all display specs"
>
> I would guess that Jan's customize fixes are responsible for
this. Stephen?
Seems unlikely to me.
Ah, but at least you replied :). Thanks for the analysis.
andy
> > xemacs,
> > - customization of Font Family of default face is not honored. I had
> > previously customized it to "Letter Gothic MT". Under 21.4.5, it is
> > displayed as "Letter Gothic MT". Under 21.4.6-rc1, it is displayed
as
> > "Courier New".
I don't have a windows build to reproduce this, however I can
reproduce the problem with 'gtk build.
In that case the problem is that customize is using font.el to do the
conversion and that does not recognize 'gtk as a device type and
silently fails.
Applying this fix makes it work
--- /local/xemacs/sunsite/xemacs-21.4/gtk/lisp/font.el~ Sun
Nov 25 17:27:33 2001
+++ /local/xemacs/sunsite/xemacs-21.4/gtk/lisp/font.el Sun
Nov 25 17:27:33 2001
＠＠ -99,6 +99,7 ＠＠
(defconst font-window-system-mappings
'((x . (x-font-create-name x-font-create-object))
+ (gtk . (x-font-create-name x-font-create-object))
(ns . (ns-font-create-name ns-font-create-object))
(mswindows . (mswindows-font-create-name
mswindows-font-create-object))
(pm . (x-font-create-name x-font-create-object)) ;
Change? FIXME
Could something similar be happening for Windows?
If so, how did customize work before this?
The cleanest way to solve this is to have make-face-size and
make-face-family use a device type specific routine a long the lines
of make-face-italic. That way the use of font.el can be dropped.
I don't think "my patch" is really causing
> When XEmacs 21.4.6rc1 is built on my system --with-gtk then default
> faces are setup correctly to GTK values, but when they are
changed with a
> subsequent M-x customize-face or custom-set-faces form the change has no
> visible effect.
either!
> If the set-face-background forms within gtk-init-device-faces aren't
> passed the gtk device as the optional locale argument (see patch) then
> everything is hunkydory and subsequent customization of the default
> faces *do* have a visible effect.
There is something else weird going on with the face frobbing. I am
going to investigate.
[time passes]
I think the problem is entirely similar to the problem we had before
with 'italic'.
Customize might need to call frob-face-property as well.
Until then,
> + ;; Removed device locales - a naive attempt to keep Custom happy.
This seems like a good idea for the moment. Gtk doesn't support
multiple device anyway AFAIK. However,
> (set-face-foreground 'highlight
> - (nth prelight (plist-get style 'text))
> - device)
> + (nth prelight (plist-get style 'text)))
> + ;device)
This not quite 100% right. It really should have been adding a '(gtk
default) tag set from the beginning, i.e. make it
(set-face-foreground 'highlight
(nth prelight (plist-get style 'text))
nil '(gtk default))
etc
Jan
11:55 p.m.
New subject: XEmacs 21.4.6-rc1 binary kit available
In that case the problem is that customize is using font.el to do
the
conversion and that does not recognize 'gtk as a device type and
silently fails.
Applying this fix makes it work
Obviously this code is already present for mswindows and used to work so
thats not it.
There is something else weird going on with the face frobbing. I am
going to investigate.
This is all gtk code so it seems unlikely that this is the cause.
andy
3:04 a.m.
New subject: XEmacs 21.4.6-rc1 binary kit available
>>>> "Jan" == Jan Vroonhof
<jan(a)xemacs.org> writes:
Jan> Applying this fix makes it work
Jan> + (gtk . (x-font-create-name x-font-create-object))
Thanks, will be in rc2.
Jan> If so, how did customize work before this?
Do you really want to hear an answer to that? :-/
Jan> That way the use of font.el can be dropped.
I know some people really don't like font.el, but having device-
specific routines without a defined API spread out over a half-dozen
files is not a very good solution either. We need some library like
font.el, if only to specify the API.
> When XEmacs 21.4.6rc1 is built on my system --with-gtk then
> default faces are setup correctly to GTK values, but when they
> are changed with a subsequent M-x customize-face or
> custom-set-faces form the change has no visible effect.
GTK support continues to be incomplete. AFAIK only Bill P understands
it at all. Bugs in the GTK support simply can't be considered
indicative of generic problems at this point.
--
Institute of Policy and Planning Sciences http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Don't ask how you can "do" free software business;
ask what your business can "do for" free software.
5:32 a.m.
New subject: XEmacs 21.4.6-rc1 binary kit available (gtk patch)
Customize works for me... I've had these patches for a while. Been
meaning to upload them. Font.el is already patched now, I've noticed.
Cheers.
edward
--- font-menu.el.orig Sun Nov 25 12:24:39 2001
+++ font-menu.el Sat Nov 24 23:39:57 2001
＠＠ -160,11 +160,13 ＠＠
(defvar font-menu-preferred-resolution
(make-specifier-and-init 'generic '((global ((mswindows) . ":")
+ ((gtk) . "*-*")
((x) . "*-*"))) t)
"Preferred horizontal and vertical font menu resolution (e.g.
\"75:75\").")
(defvar font-menu-size-scaling
(make-specifier-and-init 'integer '((global ((mswindows) . 1)
+ ((gtk) . 10)
((x) . 10))) t)
"Scale factor used in defining font sizes.")
＠＠ -285,7 +287,7 ＠＠
(entry (aref font-data 0))
(family (aref font-data 1))
(size (aref font-data 2))
- ;;(weight (aref font-data 3))
+ (weight (aref font-data 3))
s)
(unless family
(throw 'menu '(["Cannot parse current font" ding nil])))
＠＠ -316,7 +318,7 ＠＠
(font-data (font-menu-font-data 'default dcache))
(entry (aref font-data 0))
(family (aref font-data 1))
- ;;(size (aref font-data 2))
+ (size (aref font-data 2))
(weight (aref font-data 3))
w)
(unless family
＠＠ -377,7 +379,9 ＠＠
(condition-case c
(font-menu-change-face face
from-family from-weight from-size
- family weight size)
+ (or family from-family)
+ (or weight from-weight)
+ (or size from-size size))
(error
(display-error c nil)
(sit-for 1)))))
＠＠ -392,7 +396,7 ＠＠
;; OK Let Customize do it.
(custom-set-face-update-spec 'default
(list (list 'type (device-type)))
- (list :family family
+ (list :family (or family from-family)
:size (concat
(int-to-string
(/ (or size from-size)
_________________________________________________________
Do You Yahoo!?
Get your free ＠yahoo.com address at http://mail.yahoo.com
5:34 p.m.
New subject: XEmacs 21.4.6-rc1 binary kit available
At 12:04 PM 11/26/2001 +0900, Stephen J. Turnbull wrote:
I know some people really don't like font.el, but having device-
specific routines without a defined API spread out over a half-dozen
files is not a very good solution either. We need some library like
font.el, if only to specify the API.
Way back when, doing this properly in C was on the feature list for 21.4.
andy
2:39 a.m.
New subject: XEmacs 21.4.6-rc1 binary kit available
>>>> "Andy" == Andy Piper <andyp(a)bea.com>
writes:
Andy> At 12:04 PM 11/26/2001 +0900, Stephen J. Turnbull wrote:
> I know some people really don't like font.el, but having
> device- specific routines without a defined API spread out over
> a half-dozen files is not a very good solution either. We need
> some library like font.el, if only to specify the API.
Andy> Way back when, doing this properly in C
An OXYMORON. :-) C is a minimalist language for implementing minimal
functionality.
Andy> was on the feature list for 21.4.
I don't recall that. It's not to be found in Releases/Public-21.2/ on
the home page.
Certainly none of Ben, Martin, or me would have been happy about that,
on the general principle that things that can be done in Lisp should
be done in Lisp. Nor is efficiency really an excuse here;
font.el-style manipulations are only done at cus-face initialization
and user request.
Anyway, a lot of the efficiency issues could be mitigated by removing
the Customize mirroring of the information already in the faces.
Which also would be a big step toward making Customize work for the
2/3 of the world's people who don't live in ISO-8859-land. Not to
mention this maintainer. :-)
There's plenty of scope for bitflicking in redisplay. Let's get some
of those bugs and inefficiencies fixed. Many remain. :-(
All of the above is personal opinion, not Release Manager fiat, and
subject to change as I learn more. Please don't LART me yet. ;-)
--
Institute of Policy and Planning Sciences http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Don't ask how you can "do" free software business;
ask what your business can "do for" free software.
8191
days inactive
8194
days old
8 comments
5 participants
participants (5)
-
Andy Piper -
Andy Piper -
edward -
Jan Vroonhof -
Stephen J. Turnbull