10:15 a.m.
>>>> Stephen wrote:
Stephen> I really don't think those are going to be much help.
Stephen> Agreed, autoconf is a really crufty technology from the point
Stephen> of view of a build system, but its database of platform
Stephen> quirks is still unmatched AFAIK.
So it is unlikely we would change build system then due to the loss of
support for all the different platforms?
I guess our time then is best used to get autoconf working for
Xft. Could you please give a pointer to what the problem was with
getting it right? It would make it easier to help out (Huh... mabe
this is in the tracker and I haven't looked there yet.)
Yours
--
%% Mats
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
5:36 p.m.
New subject: Xft
Mats Lidell wrote:
>>>>> Stephen wrote:
Stephen> I really don't think those are going to be much help.
Stephen> Agreed, autoconf is a really crufty technology from the point
Stephen> of view of a build system, but its database of platform
Stephen> quirks is still unmatched AFAIK.
So it is unlikely we would change build system then due to the loss of
support for all the different platforms?
Is there a link/web page/something that discusses xemacs and xft? I
tried building --with-xft=emacs, and it seems to mostly work. But it
spews out lots of messages about fonts to the terminal, and the font
that it selected isn't the one I normally use (courier).
But it does look really nice, and doesn't seem to have slowed anything down.
Ray
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
6:55 p.m.
New subject: Xft
On Thu, Jul 3, 2008 at 6:36 PM, Raymond Toy (RT/EUS) <
raymond.toy(a)ericsson.com> wrote:
Mats Lidell wrote:
>>>>>> Stephen wrote:
>
> Stephen> I really don't think those are going to be much help.
> Stephen> Agreed, autoconf is a really crufty technology from the point
> Stephen> of view of a build system, but its database of platform
> Stephen> quirks is still unmatched AFAIK.
>
> So it is unlikely we would change build system then due to the loss of
> support for all the different platforms?
Is there a link/web page/something that discusses xemacs and xft? I
tried building --with-xft=emacs, and it seems to mostly work. But it
spews out lots of messages about fonts to the terminal, and the font
that it selected isn't the one I normally use (courier).
There's a node in the info file for XEmacs called 'Xft Font
Customization'
that describes how to set things up to get the right font.
Robert
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
5:11 p.m.
New subject: Xft
Robert Pluim wrote:
Is there a link/web page/something that discusses xemacs and xft?
I
tried building --with-xft=emacs, and it seems to mostly work. But it
spews out lots of messages about fonts to the terminal, and the font
that it selected isn't the one I normally use (courier).
There's a node in the info file for XEmacs called 'Xft Font
Customization' that describes how to set things up to get the right font.
How stupid of me! Sorry for the noise!
Thanks,
Ray
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
11:01 p.m.
New subject: Xft
Raymond Toy (RT/EUS) writes:
How stupid of me! Sorry for the noise!
Now that we've got you feeling guilty, maybe we can ask you where you
looked for it so we can put a cross-ref there?
Seriously, the Xft documentation was pretty poor last I worked on it.
Maybe someobody's cleaned up since I did but I don't recall seeing any
patches. Any suggestions for documentation would be appreciated (not
limited to Xft!) since people are asking for a public release. Better
doc of new features is an absolute must.
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
2:50 p.m.
New subject: Xft
Stephen J. Turnbull wrote:
Raymond Toy (RT/EUS) writes:
> How stupid of me! Sorry for the noise!
Now that we've got you feeling guilty, maybe we can ask you where you
looked for it so we can put a cross-ref there?
I did some random google searches, which didn't yield too much. I
confess I didn't try very hard.
Seriously, the Xft documentation was pretty poor last I worked on it.
Maybe someobody's cleaned up since I did but I don't recall seeing any
patches. Any suggestions for documentation would be appreciated (not
limited to Xft!) since people are asking for a public release. Better
doc of new features is an absolute must.
I think Xft is the first new feature I've wanted to try in a long time.
I will try to play around with it to see what I can get and report
back here. What I have now looks pretty good, but the font it selected
isn't very good for me because it doesn't seem to be fixed width. And
the huge stream of messages about fonts isn't very good either. That
should probably go away for the release, but ok, I guess, for betas.
On the other hand, with my new LCD display at work, xemacs without xft
looks pretty good anyway, even with my default ugly courier font. :-)
Ray
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
7:10 p.m.
New subject: Xft
Raymond Toy (RT/EUS) writes:
I did some random google searches, which didn't yield too much.
Ah. 21.5's manuals are online, I should see what "xft
site:www.xemacs.org" comes up with.
What I have now looks pretty good, but the font it selected
isn't
very good for me because it doesn't seem to be fixed width.
Aidan? Mike? You guys were the last to muck with the default list of
fonts to try.
And the huge stream of messages about fonts isn't very good
either.
You can suppress (most of) it with (setq xft-debug-level 0). If you
want to suppress all of it I guess you're going to have to edit the
DEFVAR for xft-debug-level. It's in vars_of_font_mgr() in
src/font_mgr.c. This is completely safe for you to edit, as long as
you stick to natural numbers for the value. You probably don't want
to set it to "4". :-)
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
7:12 p.m.
New subject: Xft
Stephen J. Turnbull wrote:
Raymond Toy (RT/EUS) writes:
> I did some random google searches, which didn't yield too much.
Ah. 21.5's manuals are online, I should see what "xft
site:www.xemacs.org" comes up with.
I found it now. I obviously didn't try very hard before. Sorry about that!
> What I have now looks pretty good, but the font it selected isn't
> very good for me because it doesn't seem to be fixed width.
Aidan? Mike? You guys were the last to muck with the default list of
fonts to try.
After looking over the manual, I have now been able to set the default
font to Bitstream Vera Sans Mono-11 which works well enough. (11 is a
bit small, but 12 seems overly large, and quite a big step up from 11.)
I was hoping to be able to use Adobe Courier-12 but it doesn't seem to
work. Could be a problem with my setup, but my gnome-terms use Courier
just fine.
Most of the info file is kind of useless for someone who just wants to
set a font. The sample Xresources at the very end of the section was
the most beneficial. The modeline and defaults work as specified. I
didn't compile in xft support for the menubar or tabs so I don't know if
those work or not. Still debating on whether I want to use this in my
day-to-day work.
> And the huge stream of messages about fonts isn't very good either.
You can suppress (most of) it with (setq xft-debug-level 0). If you
want to suppress all of it I guess you're going to have to edit the
DEFVAR for xft-debug-level. It's in vars_of_font_mgr() in
src/font_mgr.c. This is completely safe for you to edit, as long as
you stick to natural numbers for the value. You probably don't want
to set it to "4". :-)
Not 11? :-) I don't need to change the debug level right now, in case
there's some info there that someone might want to know.
Thanks, everyone, for your help,
Ray
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
12:50 a.m.
New subject: Xft
Mats Lidell writes:
So it is unlikely we would change build system then due to the loss
of
support for all the different platforms?
Well, somebody would have to do the research to show that a new build
system does support the main platforms (Solaris, Linux, *BSD, Mac OS
X, and Cygwin at least) well, and probably as Jerry says cobble
together a bunch of new code to support platforms/features we use but
the build system doesn't yet. Not something I want to do until we've
got a lot more manpower.
I guess our time then is best used to get autoconf working for
Xft. Could you please give a pointer to what the problem was with
getting it right?
--with-xft[=yes] does nothing; you need to use "--with-xft=[emacs,]
[menubars,][tabs,][gauges]" instead. I'm not sure this is anything
but a missing default or a documentation bug. In configure.ac, just
find the argument processing code (I think searching for "with-xft"
should do the trick). Then the code that actually processes the
arguments is an M4 macro written by me and heavily revised by Malcolm.
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
1:08 a.m.
New subject: Xft
>>>> Stephen wrote:
Stephen> --with-xft[=yes] does nothing; you need to use "--with-xft=[emacs,]
Stephen> [menubars,][tabs,][gauges]" instead. I'm not sure this is anything
Stephen> but a missing default or a documentation bug.
From looking at the source, configure and configure.ac, it seems like
yes does not accomplishes much more than making itself beeing trimmed
from the argument list used for further processing!?
The interesting part ...
----------------------------------------------------------------------
...
case "$with_xft" in
n | no | non | none ) _xft_all_default=no ;;
n,* | no,* | non,* | none,* ) _xft_all_default=no ;;
a | al | all | both ) _xft_all_default=yes ;;
a,* | al,* | all,* | both,* ) _xft_all_default=yes ;;
y | yes ) _require_xft=yes ;;
esac
if test -n "$_xft_all_default$_require_xft"; then
_with_xft_args=`echo $with_xft | sed -e 's/^[a-z]*\(,\(.*\)\)\{0,1\}$/\2/'`
else
_with_xft_args=$with_xft
fi
...
(_require_xft isn't used again until it is undefined.)
----------------------------------------------------------------------
What should "yes" mean if not equal to "all"? Why is there a yes
option at all since the docs doesn't mention it? Am I missing
something?
The same macro is used for sound, database, vdb and error_checking and
I suppose yes gives the same result there.
Yours
--
%% Mats
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
8:22 a.m.
New subject: Xft
Mats Lidell writes:
From looking at the source, configure and configure.ac, it seems
like
yes does not accomplishes much more than making itself beeing trimmed
from the argument list used for further processing!?
Actually, it makes configure use the default value.
There is a problem here: the argument default is overloaded. We need
to distinguish between the default when --with-xft isn't given at all,
and the default if the user specifies --with-xft=yes.
What should "yes" mean if not equal to "all"?
In the case of --with-xft, a plausible meaning for --with-xft=yes is
--with-xft=emacs,nomenubars,notabs,nogauges. The reason is that the
widgets don't use XEmacs's rendering engine and therefore are pretty
much restricted to ISO 8859/1. (Not to mention that gauges don't have
Xft implemented at all.)
Why is there a yes option at all since the docs doesn't mention
it?
Am I missing something?
Well, this is Malcolm's code, so he's authoritative, but I would
suppose that this is for consistency with other options. If we accept
--with-xft, and --with-xft=no, we should accept --with-xft=yes, too, I
think. The abbreviated forms for none and all aren't documented
either.
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
11:26 a.m.
New subject: Xft
>>>> Stephen wrote:
Stephen> In the case of --with-xft, a plausible meaning for
Stephen> --with-xft=yes is
Stephen> --with-xft=emacs,nomenubars,notabs,nogauges. The reason is
Stephen> that the widgets don't use XEmacs's rendering engine and
Stephen> therefore are pretty much restricted to ISO 8859/1. (Not to
Stephen> mention that gauges don't have Xft implemented at all.)
Yes, either a reasonable default or make it an error to specify
--with-xft when all the individual defaults are no.
Stephen> Well, this is Malcolm's code, so he's authoritative, but I
Stephen> would suppose that this is for consistency with other
Stephen> options. If we accept --with-xft, and --with-xft=no, we
Stephen> should accept --with-xft=yes, too, I think.
Yes, it is fine to accept it but it should mean that some reasonable
default is selected.
Yours
--
%% Mats
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
5:45 p.m.
New subject: Xft
Mats Lidell writes:
>>>>> Stephen wrote:
Stephen> Well, this is Malcolm's code, so he's
authoritative, but I
Stephen> would suppose that this is for consistency with other
Stephen> options. If we accept --with-xft, and --with-xft=no, we
Stephen> should accept --with-xft=yes, too, I think.
Yes, it is fine to accept it but it should mean that some reasonable
default is selected.
Sure. The problem is that in the case of XE_COMPLEX_OPTION, "default"
is an overloaded concept, and we'll need to fix the code to
distinguish between the "autoselected" default (which is "none") and
the default for an "Xft requested" configuration (which should not be
"none" as you point out).
For now, we could also just warn in Installation (eg, "Xft requested
but no components specified; Xft not linked.") I can do quickly that
if you thinki it would help, but I don't think it's that useful.
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
12:18 a.m.
New subject: Xft
>>>> Stephen wrote:
Stephen> For now, we could also just warn in Installation (eg, "Xft
Stephen> requested but no components specified; Xft not linked.") I
Stephen> can do quickly that if you thinki it would help, but I don't
Stephen> think it's that useful.
Why not make --with-xft=emacs,nomenubars,notabs,nogauges as you
suggest the quick fix? This will make users specifying --with-xft=yes
happier.
For the future a situations where XE_COMPLEX_OPTION has only "no" as
default, making "--with-XYZ=yes" a noop, I suggest making that a
configuration error, eg configure should complain. Just mentioning it
in Installation is not clear enough.
Yours
--
%% Mats
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
7:26 a.m.
New subject: Xft
Mats Lidell writes:
Why not make --with-xft=emacs,nomenubars,notabs,nogauges as you
suggest the quick fix? This will make users specifying --with-xft=yes
happier.
Because AFAICS it's not quick. The only way we currently have to do
that is to initialize _with_xft_emacs=yes, which will mean that
somebody who omits --with-xft entirely will be shocked when they start
XEmacs the next time. Hopefully pleasantly, but in the current state
of Xft a large minority will find it yucky to unusable.
For the future a situations where XE_COMPLEX_OPTION has only
"no" as
default, making "--with-XYZ=yes" a noop, I suggest making that a
configuration error, eg configure should complain.
Well, no, that's not really right since we can do better.
"./configure ..." (no --with-xft option at all) currently doesn't do
any Xft since I'm sufficiently unhappy with its current state
(specifically docs and the cursor turds that some fonts leave behind)
that I don't want to recommend it. This is not a bug.
"./configure --with-xft=yes" should be enough for the user to get the
"most useful, safe for a beta tester" configuration (which IMO is
"--with-xft=none,emacs"). The fact that currently it does nothing is
a bug.
Now, suppose I get fired or something nice like that so I have time to
fix the tabs to handle non-Latin-1 fonts, and I do. In that case, I
would like "--with-xft=yes" to automagically "upgrade" to
"--with-xft=none,emacs,tabs". Don't you think that is the right thing?
Just mentioning it in Installation is not clear enough.
Agreed.
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
4:21 p.m.
New subject: Xft
>>>> Stephen wrote:
Stephen> Because AFAICS it's not quick. The only way we currently have to do
Stephen> that is to initialize _with_xft_emacs=yes, which will mean that
Stephen> somebody who omits --with-xft entirely will be shocked when they start
Stephen> XEmacs the next time. Hopefully pleasantly, but in the current state
Stephen> of Xft a large minority will find it yucky to unusable.
OOPS! Forget all I said. We don't want that.
I thought that the defaults only was applied when --with-XYZ was
specified not that they were _the_ defaults. Sorry.
Stephen> Well, no, that's not really right since we can do better.
Fine. So XE_COMPLEX_OPTION needs two defaults then. One default when
the options isn't mentioned at all and another default when the user
wants the option but with no details specified? This doesn't sound
that hard to get right.
Maybe XE_COMPLEX_OPTION works for the other options except xft so we
ought to introduce another macro XE_XFT_OPTION in order to play with
it without messing up for other options?
Yours
--
%% Mats
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
7:18 p.m.
New subject: Xft
Mats Lidell writes:
>>>>> Stephen wrote:
Stephen> Well, no, that's not really right since we can do
better.
Fine. So XE_COMPLEX_OPTION needs two defaults then. One default when
the options isn't mentioned at all and another default when the user
wants the option but with no details specified? This doesn't sound
that hard to get right.
No. It's just that Malcolm completely rewrote my code so I don't know
how to do it without studying his.
Maybe XE_COMPLEX_OPTION works for the other options except xft so we
ought to introduce another macro XE_XFT_OPTION in order to play with
it without messing up for other options?
I don't think so. I mean, we all do keep a reasonably satisfactory
copy of XEmacs installed somewhere, right? So it's only the bleeding
edge tester version that might lose some desired options. Just be
moderately careful to make sure that an explicit spec for all
components works correctly.
If you feel really nervous, just tell Mike and Vin *not* to move your
patch from xemacs branch to xemacs-beta branch until you say so.
[Aside to Mike and Vin: I think you *should* accept such requests for
now, but we need to be careful about the workflow so it doesn't become
a burden on you.]
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
8:41 p.m.
New subject: Xft
"Stephen J. Turnbull" <stephen(a)xemacs.org> writes:
[Aside to Mike and Vin: I think you *should* accept such requests
for
now, but we need to be careful about the workflow so it doesn't become
a burden on you.]
So far, it's no burden at all. It'd be a nice change if something
interesting happened :-)
--
Cheers =8-} Mike
Friede, Völkerverständigung und überhaupt blabla
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
1:57 p.m.
New subject: Xft
>>>> Stephen wrote:
>> >>>> Stephen wrote:
Stephen> Well, no, that's not really right since we can do better.
>
> Fine. So XE_COMPLEX_OPTION needs two defaults then. One default when
> the options isn't mentioned at all and another default when the user
> wants the option but with no details specified? This doesn't sound
> that hard to get right.
Stephen> No. It's just that Malcolm completely rewrote my code so I don't
know
Stephen> how to do it without studying his.
Well it seems like the configure options together with macro language
m4 sooner or later gets you into a twisty little maze of passages
... ;-) Maybe there is a way out somewhere but I haven't found it yet!
Anyway for a little patchy, quick!?, solution this might work. Change
the default to what is wanted when selected and set it back to none
selected in the else path.
--- working directory: "/src/xemacs/xft/"
% hg diff configure.ac
diff -r ec442dc06fe1 configure.ac
--- a/configure.ac Thu Jul 10 23:37:52 2008 +0200
+++ b/configure.ac Sun Jul 13 14:33:48 2008 +0200
＠＠ -690,9 +690,10 ＠＠
[Xft],
[`emacs' (buffers), `menubars', `tabs', and `gauges'],
[X11, Xft, Xrender, freetype, and fontconfig],
- [`noemacs, nomenubars, notabs, nogauges'])],
- [],[],
- [XE_COMPLEX_OPTION([emacs],[no]),
+ [`noemacs, nomenubars, notabs, nogauges'.
+ The default when selected is `emacs, nomenubars, notabs, nogauges'])],
+ [],[with_xft_emacs=no],
+ [XE_COMPLEX_OPTION([emacs],[yes]),
XE_COMPLEX_OPTION([menubars],[no]),
XE_COMPLEX_OPTION([tabs],[no]),
XE_COMPLEX_OPTION([gauges],[no])])
Yours
--
%% Mats
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
5760
days inactive
5770
days old
18 comments
6 participants
participants (6)
-
Michael Sperber -
Raymond Toy -
Robert Pluim -
Stephen J. Turnbull -
Stephen J. Turnbull -
The XEmacs Package Smoketest