Emacs 23.2 plans

[Chong Yidong]

Stefan and I plan to begin the Emacs 23.2 pretest in the beginning of
December.  Accordingly, we will be imposing a freeze on new features
(i.e., anything that is not a bugfix) within the next few weeks.  The
target for Emacs 23.2 is March next year; as usual, the actual release
will depend on how pretesting goes.

If you are planning to make any significant changes to the tree, please
let us know ASAP (i.e. now), so that we can plan/comment accordingly.

[Giorgos Keramidas]

On Sat, 07 Nov 2009 23:46:44 -0500, Chong Yidong <cyd@...> wrote:
> Stefan and I plan to begin the Emacs 23.2 pretest in the beginning of
> December.  Accordingly, we will be imposing a freeze on new features
> (i.e., anything that is not a bugfix) within the next few weeks.  The
> target for Emacs 23.2 is March next year; as usual, the actual release
> will depend on how pretesting goes.
>
> If you are planning to make any significant changes to the tree, please
> let us know ASAP (i.e. now), so that we can plan/comment accordingly.

Are documentation changes ok during the final release stages?

I think my copyright assignment papers may be ok by then, so I keep
typing a smallish section for `IBuffer' to attach to our current "Buffer
Convenience" node of doc/emacs/buffers.texi.

[Dan Nicolaescu]

Chong Yidong <cyd@...> writes:

  > Stefan and I plan to begin the Emacs 23.2 pretest in the beginning of
  > December.  Accordingly, we will be imposing a freeze on new features
  > (i.e., anything that is not a bugfix) within the next few weeks.  The
  > target for Emacs 23.2 is March next year; as usual, the actual release
  > will depend on how pretesting goes.
  >
  > If you are planning to make any significant changes to the tree, please
  > let us know ASAP (i.e. now), so that we can plan/comment accordingly.

I have a few more patches that purecopy various things that are not
checked in yet (they just need a ChangeLog).

I would also like to get rid of the platforms that were obsoleted in
etc/NEWS in 23.1.  May I do that ASAP?

Also there are quite a few bugs in the bug tracker that are tagged with
"patch" (and some that are not, but still contain patches).  It would be
nice if some/most of those could be checked in.

[Stefan Monnier]

> Are documentation changes ok during the final release stages?

Yes, documentation changes can be applied pretty much at any time.


        Stefan

[Carsten Dominik-3]

On Nov 8, 2009, at 5:46 AM, Chong Yidong wrote:

> Stefan and I plan to begin the Emacs 23.2 pretest in the beginning of
> December.  Accordingly, we will be imposing a freeze on new features
> (i.e., anything that is not a bugfix) within the next few weeks.  The
> target for Emacs 23.2 is March next year; as usual, the actual release
> will depend on how pretesting goes.
>
> If you are planning to make any significant changes to the tree,  
> please
> let us know ASAP (i.e. now), so that we can plan/comment accordingly.

I have a major chunk of org-mode changes coming in, by the
end of the week, hopefully.

- Carsten

[Alan Mackenzie]

Hi, Yidong,

On Sat, Nov 07, 2009 at 11:46:44PM -0500, Chong Yidong wrote:
> Stefan and I plan to begin the Emacs 23.2 pretest in the beginning of
> December.  Accordingly, we will be imposing a freeze on new features
> (i.e., anything that is not a bugfix) within the next few weeks.  The
> target for Emacs 23.2 is March next year; as usual, the actual release
> will depend on how pretesting goes.

> If you are planning to make any significant changes to the tree, please
> let us know ASAP (i.e. now), so that we can plan/comment accordingly.

Yes.  I am enhancing CC Mode's function `c-parse-state' so that it can
handle "brace deserts" (source files with few or no braces) at a
tolerable speed.  Currently, (in a file supplied by Dave Milter, who
reported the bug), scrolling back a screen from EOB takes several
seconds.

This change involves ~1200 new lines of code in cc-engine.el, and is
about "99% complete".  I anticipate commiting it in a few days time,
before the end of the month.

--
Alan Mackenzie (Nuremberg, Germany).

[Sam Steingold]

Alan Mackenzie wrote:
> I am enhancing CC Mode's function `c-parse-state' so that it can
> handle "brace deserts" (source files with few or no braces) at a
> tolerable speed.  Currently, (in a file supplied by Dave Milter, who
> reported the bug), scrolling back a screen from EOB takes several
> seconds.

well, I have a whole bunch of speed test files for you: clisp sources.
would you like to take a look?

[Jan Djärv]

Chong Yidong skrev:
> Stefan and I plan to begin the Emacs 23.2 pretest in the beginning of
> December.  Accordingly, we will be imposing a freeze on new features
> (i.e., anything that is not a bugfix) within the next few weeks.  The
> target for Emacs 23.2 is March next year; as usual, the actual release
> will depend on how pretesting goes.
>
> If you are planning to make any significant changes to the tree, please
> let us know ASAP (i.e. now), so that we can plan/comment accordingly.
>

I have completed the GConf/Xsetting code (after way to long time) now and
would like to check it in.
I don't know if it counts as a major change, but its implications might.

Basically Emacs reads the system font name from GConf and uses that.  It also
reacts to changes to the system font (can be turned off with a defcustom).
GConf support can be turned off in configure.

Emacs also listens for font parameter changes made by the user and
communicated with Xsettings.  These are various Xft-related parameters
(antialias, hinting, hintstyle, RGBA, DPI and lcdfilter) that are settable by
the Gnome user settings program (I imagine KDE has something similar).

The upside is that Emacs integrates better to the Gnome desktop and that we
can control Xft-settings.  After all, if a user changes hinting she probably
wants it for all applications.  As Emacs now uses the DPI value from Xsettings
instead of from the X server, fonst look the same.  This isn't the case now,
Monospace-8 on Emacs and gnome-terminal look quite different.

The downside is that user might be surprised to see the size of their fonts
change as the DPI value is now different.

What do you think?

        Jan D.

[Chong Yidong]

Jan Djärv <jan.h.d@...> writes:

> I have completed the GConf/Xsetting code (after way to long time) now
> and would like to check it in.  I don't know if it counts as a major
> change, but its implications might.
>
> Basically Emacs reads the system font name from GConf and uses that.
> It also reacts to changes to the system font (can be turned off with a
> defcustom). GConf support can be turned off in configure.

I'm not sure if there's enough time to test this.  Font-related changes
seem pretty tricky to get right.  What about checking this in but
disabling the configure option by default, and enabling it after Emacs
23.2?

[Stefan Monnier]

> The upside is that Emacs integrates better to the Gnome desktop and
> that we can control Xft-settings.  After all, if a user changes
> hinting she probably wants it for all applications.  As Emacs now uses
> the DPI value from Xsettings instead of from the X server, fonst look
> the same.  This isn't the case now, Monospace-8 on Emacs and
> gnome-terminal look quite different.

That's a very good change, yes.

> The downside is that user might be surprised to see the size of their fonts
> change as the DPI value is now different.

If it can fix the problem that the default font size is much too large
for all my screens (a 1600x1200 21" desktop, a 1400x1050 14" laptop,
and a 1024x600 10" netbook), then I'm all for it, and I know I wouldn't
be the only user to appreciate the change.  See bug#4547 for example.


        Stefan

[Jan Djärv]

Chong Yidong skrev:

> Jan Djärv <jan.h.d@...> writes:
>
>> I have completed the GConf/Xsetting code (after way to long time) now
>> and would like to check it in.  I don't know if it counts as a major
>> change, but its implications might.
>>
>> Basically Emacs reads the system font name from GConf and uses that.
>> It also reacts to changes to the system font (can be turned off with a
>> defcustom). GConf support can be turned off in configure.
>
> I'm not sure if there's enough time to test this.  Font-related changes
> seem pretty tricky to get right.  What about checking this in but
> disabling the configure option by default, and enabling it after Emacs
> 23.2?

I've done so.  But IMHO a couple of monthe pre-test is time enough to test it.

        Jan D.

[Tobias C. Rittweiler]

Jan Djärv <jan.h.d@...> writes:

> Chong Yidong skrev:
>> Jan Djärv <jan.h.d@...> writes:
>>
>>> I have completed the GConf/Xsetting code (after way to long time) now
>>> and would like to check it in.  I don't know if it counts as a major
>>> change, but its implications might.
>>>
>>> Basically Emacs reads the system font name from GConf and uses that.
>>> It also reacts to changes to the system font (can be turned off with a
>>> defcustom). GConf support can be turned off in configure.
>>
>> I'm not sure if there's enough time to test this.  Font-related changes
>> seem pretty tricky to get right.  What about checking this in but
>> disabling the configure option by default, and enabling it after Emacs
>> 23.2?
>
> I've done so.  But IMHO a couple of monthe pre-test is time enough to test it.

I'm interested in the change. How can I test it?

  -T.

[Jan Djärv]

Tobias C. Rittweiler skrev:

> Jan Djärv <jan.h.d@...> writes:
>
>> Chong Yidong skrev:
>>> Jan Djärv <jan.h.d@...> writes:
>>>
>>>> I have completed the GConf/Xsetting code (after way to long time) now
>>>> and would like to check it in.  I don't know if it counts as a major
>>>> change, but its implications might.
>>>>
>>>> Basically Emacs reads the system font name from GConf and uses that.
>>>> It also reacts to changes to the system font (can be turned off with a
>>>> defcustom). GConf support can be turned off in configure.
>>> I'm not sure if there's enough time to test this.  Font-related changes
>>> seem pretty tricky to get right.  What about checking this in but
>>> disabling the configure option by default, and enabling it after Emacs
>>> 23.2?
>> I've done so.  But IMHO a couple of monthe pre-test is time enough to test it.
>
> I'm interested in the change. How can I test it?
>

Make sure you have gconf development libraries installed (libgconf(2)-dev(el)
or similar package).
Run configure.  Set font-use-system-font to t (it is in the menu Options also).
Use Gnome settings to change the monospaced font.
Also try to change dpi, hinting style and so on (this should work on KDE
also). Emacs frames should be updated at once.

        Jan D.

[Jan Djärv]

Stefan Monnier skrev:

>> The upside is that Emacs integrates better to the Gnome desktop and
>> that we can control Xft-settings.  After all, if a user changes
>> hinting she probably wants it for all applications.  As Emacs now uses
>> the DPI value from Xsettings instead of from the X server, fonst look
>> the same.  This isn't the case now, Monospace-8 on Emacs and
>> gnome-terminal look quite different.
>
> That's a very good change, yes.
>
>> The downside is that user might be surprised to see the size of their fonts
>> change as the DPI value is now different.
>
> If it can fix the problem that the default font size is much too large
> for all my screens (a 1600x1200 21" desktop, a 1400x1050 14" laptop,
> and a 1024x600 10" netbook), then I'm all for it, and I know I wouldn't
> be the only user to appreciate the change.  See bug#4547 for example.
>

I will add some comments to that bug. It is confusing that DPI isn't really
the physical DPI, but I think we should just adapt and move on.  Can you try
Emacs now and see if it fixes your problem? It would make a good verification
test.

        Jan D.

[YAMAMOTO Mitsuharu]

>>>>> On Tue, 17 Nov 2009 09:31:39 +0100, Jan Djärv <jan.h.d@...> said:

>> I'm not sure if there's enough time to test this.  Font-related
>> changes seem pretty tricky to get right.  What about checking this
>> in but disabling the configure option by default, and enabling it
>> after Emacs 23.2?

> I've done so.  But IMHO a couple of monthe pre-test is time enough
> to test it.

It seems that FC_LCD_* used in xftfont.c and xsettings.c are not
defined in /usr/X11/include/fontconfig/fontconfig.h on Mac OS X
10.5.8, which says:

#define FC_MAJOR 2
#define FC_MINOR 4
#define FC_REVISION 2

Perhaps some fallback definitions for FC_LCD_* can be provided
somewhere so that a binary compiled on an older system can use the LCD
features when it runs on a newer one.

                                     YAMAMOTO Mitsuharu
                                mituharu@...

[Jan Djärv]

YAMAMOTO Mitsuharu skrev:

>>>>>> On Tue, 17 Nov 2009 09:31:39 +0100, Jan Djärv <jan.h.d@...> said:
>
>>> I'm not sure if there's enough time to test this.  Font-related
>>> changes seem pretty tricky to get right.  What about checking this
>>> in but disabling the configure option by default, and enabling it
>>> after Emacs 23.2?
>
>> I've done so.  But IMHO a couple of monthe pre-test is time enough
>> to test it.
>
> It seems that FC_LCD_* used in xftfont.c and xsettings.c are not
> defined in /usr/X11/include/fontconfig/fontconfig.h on Mac OS X
> 10.5.8, which says:
>
> #define FC_MAJOR 2
> #define FC_MINOR 4
> #define FC_REVISION 2
>
> Perhaps some fallback definitions for FC_LCD_* can be provided
> somewhere so that a binary compiled on an older system can use the LCD
> features when it runs on a newer one.
>

I added an #ifdef for FC_LCD_FILTER.  If I add fallbacks, the older version
won't work.  It will refuse to add keywords (lcdfilter) it doesn't know about
and return an error.

        Jan D.

[YAMAMOTO Mitsuharu]

>>>>> On Tue, 17 Nov 2009 10:58:56 +0100, Jan Djärv <jan.h.d@...> said:

>> Perhaps some fallback definitions for FC_LCD_* can be provided
>> somewhere so that a binary compiled on an older system can use the
>> LCD features when it runs on a newer one.
>>

> I added an #ifdef for FC_LCD_FILTER.  If I add fallbacks, the older
> version won't work.  It will refuse to add keywords (lcdfilter) it
> doesn't know about and return an error.

Hmm, that's unfortunate...

                                     YAMAMOTO Mitsuharu
                                mituharu@...

[Tobias C. Rittweiler]

Jan Djärv <jan.h.d@...> writes:

> Tobias C. Rittweiler skrev:
>> Jan Djärv <jan.h.d@...> writes:
>>
>>> Chong Yidong skrev:
>>>> Jan Djärv <jan.h.d@...> writes:
>>>>
>>>>> I have completed the GConf/Xsetting code (after way to long time) now
>>>>> and would like to check it in.  I don't know if it counts as a major
>>>>> change, but its implications might.
>>>>>
>>>>> Basically Emacs reads the system font name from GConf and uses that.
>>>>> It also reacts to changes to the system font (can be turned off with a
>>>>> defcustom). GConf support can be turned off in configure.
>>>> I'm not sure if there's enough time to test this.  Font-related changes
>>>> seem pretty tricky to get right.  What about checking this in but
>>>> disabling the configure option by default, and enabling it after Emacs
>>>> 23.2?
>>> I've done so.  But IMHO a couple of monthe pre-test is time enough to test it.
>>
>> I'm interested in the change. How can I test it?
>>
>
> Make sure you have gconf development libraries installed
> (libgconf(2)-dev(el) or similar package).
> Run configure.  Set font-use-system-font to t (it is in the menu Options also).
> Use Gnome settings to change the monospaced font.
> Also try to change dpi, hinting style and so on (this should work on
> KDE also). Emacs frames should be updated at once.

Trying to compile with

  ./configure --prefix=... --with-x-toolkit=gtk  --with-xpm=n

results in the following build error:

  gcc -c -D_BSD_SOURCE   -Demacs -DHAVE_CONFIG_H  -I. -I/home/tcr/src/from-upstream/emacs/src -D_BSD_SOURCE -I/usr/include/gtk-2.0 -I/usr/lib/gtk-2.0/include -I/usr/include/atk-1.0 -I/usr/include/cairo -I/usr/include/pango-1.0 -I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include -I/usr/include/freetype2 -I/usr/include/libpng12 -I/usr/include/pixman-1 -I/usr/include/freetype2   -I/usr/include/dbus-1.0 -I/usr/lib/dbus-1.0/include -DORBIT2=1 -pthread -I/usr/include/gconf/2 -I/usr/include/orbit-2.0 -I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include -g -O2 -Wno-pointer-sign   -MMD -MF deps/xsettings.d xsettings.c
  xsettings.c: In function ‘parse_xft_settings’:
  xsettings.c:318: error: ‘FC_LCD_NONE’ undeclared (first use in this function)
  xsettings.c:318: error: (Each undeclared identifier is reported only once
  xsettings.c:318: error: for each function it appears in.)
  xsettings.c:320: error: ‘FC_LCD_DEFAULT’ undeclared (first use in this function)
  make[1]: *** [xsettings.o] Error 1
  make[1]: Leaving directory `/src/tcr/from-upstream/emacs/src'
  make: *** [src] Error 2

The topmost entry in src/ChangeLog is

  2009-11-17  Jan Djärv  <jan.h.d@...>
 
          * xftfont.c (xftfont_fix_match): Older versions of fontconfig does
          not have FC_LCD_FILTER.  #ifdef it.

         ...

This is

  Linux thaleron 2.6.24-25-386 #1 Tue Oct 20 06:50:20 UTC 2009 i686
  GNU/Linux

The version of my libgconf2-dev package is: 2.22.0-ubuntu3
                  libxft/-dev   package is; 2.1.12-2ubuntu5

Anything else of interest?

  -T.

[Jan Djärv]

YAMAMOTO Mitsuharu skrev:

>>>>>> On Tue, 17 Nov 2009 10:58:56 +0100, Jan Djärv <jan.h.d@...> said:
>
>>> Perhaps some fallback definitions for FC_LCD_* can be provided
>>> somewhere so that a binary compiled on an older system can use the
>>> LCD features when it runs on a newer one.
>>>
>
>> I added an #ifdef for FC_LCD_FILTER.  If I add fallbacks, the older
>> version won't work.  It will refuse to add keywords (lcdfilter) it
>> doesn't know about and return an error.
>
> Hmm, that's unfortunate...
>

Actually, I think i misread the code.  I'll have another look and add
fallbacks if it will work.

        Jan D.

[Jan Djärv]

Tobias C. Rittweiler skrev:

> Jan Djärv <jan.h.d@...> writes:
>
>> Tobias C. Rittweiler skrev:
>>> Jan Djärv <jan.h.d@...> writes:
>>>
>>>> Chong Yidong skrev:
>>>>> Jan Djärv <jan.h.d@...> writes:
>>>>>
>>>>>> I have completed the GConf/Xsetting code (after way to long time) now
>>>>>> and would like to check it in.  I don't know if it counts as a major
>>>>>> change, but its implications might.
>>>>>>
>>>>>> Basically Emacs reads the system font name from GConf and uses that.
>>>>>> It also reacts to changes to the system font (can be turned off with a
>>>>>> defcustom). GConf support can be turned off in configure.
>>>>> I'm not sure if there's enough time to test this.  Font-related changes
>>>>> seem pretty tricky to get right.  What about checking this in but
>>>>> disabling the configure option by default, and enabling it after Emacs
>>>>> 23.2?
>>>> I've done so.  But IMHO a couple of monthe pre-test is time enough to test it.
>>> I'm interested in the change. How can I test it?
>>>
>> Make sure you have gconf development libraries installed
>> (libgconf(2)-dev(el) or similar package).
>> Run configure.  Set font-use-system-font to t (it is in the menu Options also).
>> Use Gnome settings to change the monospaced font.
>> Also try to change dpi, hinting style and so on (this should work on
>> KDE also). Emacs frames should be updated at once.
>
> Trying to compile with
>
>   ./configure --prefix=... --with-x-toolkit=gtk  --with-xpm=n
>
> results in the following build error:
>
>   gcc -c -D_BSD_SOURCE   -Demacs -DHAVE_CONFIG_H  -I. -I/home/tcr/src/from-upstream/emacs/src -D_BSD_SOURCE -I/usr/include/gtk-2.0 -I/usr/lib/gtk-2.0/include -I/usr/include/atk-1.0 -I/usr/include/cairo -I/usr/include/pango-1.0 -I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include -I/usr/include/freetype2 -I/usr/include/libpng12 -I/usr/include/pixman-1 -I/usr/include/freetype2   -I/usr/include/dbus-1.0 -I/usr/lib/dbus-1.0/include -DORBIT2=1 -pthread -I/usr/include/gconf/2 -I/usr/include/orbit-2.0 -I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include -g -O2 -Wno-pointer-sign   -MMD -MF deps/xsettings.d xsettings.c
>   xsettings.c: In function ‘parse_xft_settings’:
>   xsettings.c:318: error: ‘FC_LCD_NONE’ undeclared (first use in this function)
>   xsettings.c:318: error: (Each undeclared identifier is reported only once
>   xsettings.c:318: error: for each function it appears in.)
>   xsettings.c:320: error: ‘FC_LCD_DEFAULT’ undeclared (first use in this function)
>   make[1]: *** [xsettings.o] Error 1
>   make[1]: Leaving directory `/src/tcr/from-upstream/emacs/src'
>   make: *** [src] Error 2
>

Interesting, you have FC_LCD_FILTER but not the others.  Please try again.

        Jan D.