Preload encoded-kb on MS-Windows

Preload encoded-kb on MS-Windows

[Eli Zaretskii]

On MS-Windows, the startup code always causes encoded-kb to be loaded,
because of this snippet from mule-cmds.el

    ;; On Windows, override locale-coding-system,
    ;; keyboard-coding-system with system codepage.  Note:
    ;; selection-coding-system is already set in w32select.c.
    (when (boundp 'w32-ansi-code-page)
      (let ((code-page-coding (intern (format "cp%d" w32-ansi-code-page))))
	(when (coding-system-p code-page-coding)
	  (setq locale-coding-system code-page-coding)
	  (set-keyboard-coding-system code-page-coding)
	  (set-terminal-coding-system code-page-coding))))

This calls set-keyboard-coding-system, which in turn loads encoded-kb
to enable Encoded Keyboard minor mode.

I think this means we should preload encoded-kb on Windows.  Does
anyone disagree?

Re: Preload encoded-kb on MS-Windows

[Stefan Monnier]

>     ;; On Windows, override locale-coding-system,
>     ;; keyboard-coding-system with system codepage.  Note:
>     ;; selection-coding-system is already set in w32select.c.
>     (when (boundp 'w32-ansi-code-page)
>       (let ((code-page-coding (intern (format "cp%d" w32-ansi-code-page))))
> 	(when (coding-system-p code-page-coding)
> 	  (setq locale-coding-system code-page-coding)
> 	  (set-keyboard-coding-system code-page-coding)
> 	  (set-terminal-coding-system code-page-coding))))

> This calls set-keyboard-coding-system, which in turn loads encoded-kb
> to enable Encoded Keyboard minor mode.

> I think this means we should preload encoded-kb on Windows.  Does
> anyone disagree?

Actually under Unix, unless your locale is C or English,
terminal-coding-system will most likely be set as well.
Under Mac OS X, it's always set to utf-8.
So we should probably just preload it in all cases.


        Stefan

Re: Preload encoded-kb on MS-Windows

[Kenichi Handa]

In article <873bfds4xs.fsf-monnier+emacs@gnu.org>, Stefan Monnier <monnier@iro.umontreal.ca> writes:

>> ;; On Windows, override locale-coding-system,
>> ;; keyboard-coding-system with system codepage.  Note:
>> ;; selection-coding-system is already set in w32select.c.
>> (when (boundp 'w32-ansi-code-page)
>> (let ((code-page-coding (intern (format "cp%d" w32-ansi-code-page))))
>> (when (coding-system-p code-page-coding)
>> (setq locale-coding-system code-page-coding)
>> (set-keyboard-coding-system code-page-coding)
>> (set-terminal-coding-system code-page-coding))))

>> This calls set-keyboard-coding-system, which in turn loads encoded-kb
>> to enable Encoded Keyboard minor mode.

>> I think this means we should preload encoded-kb on Windows.  Does
>> anyone disagree?

> Actually under Unix, unless your locale is C or English,
> terminal-coding-system will most likely be set as well.
> Under Mac OS X, it's always set to utf-8.
> So we should probably just preload it in all cases.

It seems that you actually meant keyboard-coding-system (not
terminal-coding-system).  But, on Unix, encoded-kb is not
loaded when you run Emacs on X.

---
Kenichi Handa
handa@m17n.org

Re: Preload encoded-kb on MS-Windows

[Eli Zaretskii]

> From: Kenichi Handa <handa@m17n.org>
> CC: eliz@gnu.org, emacs-devel@gnu.org
> Date: Thu, 18 May 2006 10:20:38 +0900
> 
> It seems that you actually meant keyboard-coding-system (not
> terminal-coding-system).  But, on Unix, encoded-kb is not
> loaded when you run Emacs on X.

Do you agree that on MS-Windows encoded-kb should be preloaded?

Re: Preload encoded-kb on MS-Windows

[Kenichi Handa]

In article <u4pzo2e6t.fsf@gnu.org>, Eli Zaretskii <eliz@gnu.org> writes:

>> From: Kenichi Handa <handa@m17n.org>
>> CC: eliz@gnu.org, emacs-devel@gnu.org
>> Date: Thu, 18 May 2006 10:20:38 +0900
>> 
>> It seems that you actually meant keyboard-coding-system (not
>> terminal-coding-system).  But, on Unix, encoded-kb is not
>> loaded when you run Emacs on X.

> Do you agree that on MS-Windows encoded-kb should be preloaded?

I agree if it is the policy to preload everything that is
always loaded at startup time.

But, as a developper, I see a merit in not preloading such
files because then the debugging/testing of them gets a
little bit easier (we don't have to dump Emacs).

---
Kenichi Handa
handa@m17n.org

Re: Preload encoded-kb on MS-Windows

[Eli Zaretskii]

> From: Kenichi Handa <handa@m17n.org>
> CC: monnier@iro.umontreal.ca, emacs-devel@gnu.org
> Date: Thu, 18 May 2006 12:55:16 +0900
> 
> In article <u4pzo2e6t.fsf@gnu.org>, Eli Zaretskii <eliz@gnu.org> writes:
> 
> >> From: Kenichi Handa <handa@m17n.org>
> >> CC: eliz@gnu.org, emacs-devel@gnu.org
> >> Date: Thu, 18 May 2006 10:20:38 +0900
> >> 
> >> It seems that you actually meant keyboard-coding-system (not
> >> terminal-coding-system).  But, on Unix, encoded-kb is not
> >> loaded when you run Emacs on X.
> 
> > Do you agree that on MS-Windows encoded-kb should be preloaded?
> 
> I agree if it is the policy to preload everything that is
> always loaded at startup time.

I thought this was the policy, because it makes Emacs start faster.
Richard, is it true that this is the policy?

> But, as a developper, I see a merit in not preloading such
> files because then the debugging/testing of them gets a
> little bit easier (we don't have to dump Emacs).

Well, we have gobs of files preloaded already, so one more won't hurt
too much, I think.  But I'll defer to Richard's decision.

Re: Preload encoded-kb on MS-Windows

[Kenichi Handa]

In article <u3bf72mje.fsf@gnu.org>, Eli Zaretskii <eliz@gnu.org> writes:

>> But, as a developper, I see a merit in not preloading such
>> files because then the debugging/testing of them gets a
>> little bit easier (we don't have to dump Emacs).

> Well, we have gobs of files preloaded already, so one more won't hurt
> too much, I think.  But I'll defer to Richard's decision.

It seems that you misunderstand my point above.  When *.el
is not preloaded, we can test it by modifying and
byte-compiling that file and just restart Emacs.  If *.el is
preloaded, we need an additional step; building Emacs.

---
Kenichi Handa
handa@m17n.org

Re: Preload encoded-kb on MS-Windows

[Eli Zaretskii]

> From: Kenichi Handa <handa@m17n.org>
> CC: monnier@iro.umontreal.ca, emacs-devel@gnu.org
> Date: Fri, 19 May 2006 09:33:21 +0900
> 
> In article <u3bf72mje.fsf@gnu.org>, Eli Zaretskii <eliz@gnu.org> writes:
> 
> >> But, as a developper, I see a merit in not preloading such
> >> files because then the debugging/testing of them gets a
> >> little bit easier (we don't have to dump Emacs).
> 
> > Well, we have gobs of files preloaded already, so one more won't hurt
> > too much, I think.  But I'll defer to Richard's decision.
> 
> It seems that you misunderstand my point above.  When *.el
> is not preloaded, we can test it by modifying and
> byte-compiling that file and just restart Emacs.  If *.el is
> preloaded, we need an additional step; building Emacs.

I did understand what you were saying.  My response to that is that we
already have that situation with many other files: e.g., when we need
to debug files.el, we need to rebuild Emacs.

Re: Preload encoded-kb on MS-Windows

[Kenichi Handa]

In article <uk68i1fpb.fsf@gnu.org>, Eli Zaretskii <eliz@gnu.org> writes:

> I did understand what you were saying.  My response to that is that we
> already have that situation with many other files: e.g., when we need
> to debug files.el, we need to rebuild Emacs.

Oops, sorry for the misunderstanding on my side.

---
Kenichi Handa
handa@m17n.org

Re: Preload encoded-kb on MS-Windows

[Richard Stallman]

    > I agree if it is the policy to preload everything that is
    > always loaded at startup time.

    I thought this was the policy, because it makes Emacs start faster.
    Richard, is it true that this is the policy?

Yes, unless there is some specific pressing reason to do otherwise.