Carbon port emacs-unicode-2 build problem under MacOSX

* Carbon port emacs-unicode-2 build problem under MacOSX
@ 2007-11-06 12:02 CHENG Gao
  2007-11-06 12:14 ` CHENG Gao
                   ` (3 more replies)
  0 siblings, 4 replies; 44+ messages in thread
From: CHENG Gao @ 2007-11-06 12:02 UTC (permalink / raw)
  To: emacs-devel

My last successfully build was on May 26. Yesterday I gave it another
try and found something.

I reported twice before about failing owing to undefined symbols in
macterm.c. The problem is from new mac_set_unicode_keystroke_event
introduced by June 7 checkin (revision 1.47.2.56, Wed Jun 7 18:04:51
2006 UTC).

The code:
,----
| static void
| 	 	 mac_set_unicode_keystroke_event (code, buf)
| 	 	      UniChar code;
| 	 	      struct input_event *buf;
| 	 	 {
| 	 	   int charset_id, c1, c2;
| 	 	 
| 	 	   if (code < 0x80)
| 	 	     {
| 	 	       buf->kind = ASCII_KEYSTROKE_EVENT;
| 	 	       buf->code = code;
| 	 	     }
| 	 	   else if (code < 0x100)
| 	 	     {
| 	 	       if (code < 0xA0)
| 	 	         charset_id = CHARSET_8_BIT_CONTROL;
| 	 	       else
| 	 	         charset_id = charset_latin_iso8859_1;
| 	 	       buf->kind = MULTIBYTE_CHAR_KEYSTROKE_EVENT;
| 	 	       buf->code = MAKE_CHAR (charset_id, code, 0);
| 	 	     }
| 	 	   else
| 	 	     {
| 	 	       if (code < 0x2500)
| 	 	         charset_id = charset_mule_unicode_0100_24ff,
| 	 	           code -= 0x100;
| 	 	       else if (code < 0x33FF)
| 	 	         charset_id = charset_mule_unicode_2500_33ff,
| 	 	           code -= 0x2500;
| 	 	       else if (code >= 0xE000)
| 	 	         charset_id = charset_mule_unicode_e000_ffff,
| 	 	           code -= 0xE000;
| 	 	       c1 = (code / 96) + 32, c2 = (code % 96) + 32;
| 	 	       buf->kind = MULTIBYTE_CHAR_KEYSTROKE_EVENT;
| 	 	       buf->code = MAKE_CHAR (charset_id, c1, c2);
| 	 	     }
| 	 	 }
`----

CHARSET_8_BIT_CONTROL, charset_mule_unicode_0100_24ff,
charset_mule_2500_33ff, charset_mule_unicode_e000_ffff used here are
defined in any place.

I reverted this funciton to previous as:
,----
| static void
| mac_set_unicode_keystroke_event (code, buf)
|      UniChar code;
|      struct input_event *buf;
| {
|   int charset_id, c1, c2;
| 
|   if (code < 0x80)
|     buf->kind = ASCII_KEYSTROKE_EVENT;
|   else
|     buf->kind = MULTIBYTE_CHAR_KEYSTROKE_EVENT;
|   buf->code = code;
| }
`----

Now macterm.c can be built.

Another error (for building latest cvs source) is "Undefined symbols:
_res_init_9". I googled and found a workaround, that's do

,----
| export LIBS=-lresolv
`----

before "make bootstrap".
(Original URL:
http://groups.google.com.ua/group/gnu.emacs.bug/msg/70a2759b8d6069e1 )

Thus I can build emacs-unicode-2 successfully.

Emacs running from terminal works ok. But Emacs.app (running from
Finder) is not usable. For every keystroke (keyboard input or mouse
click) I have to wait for several minutes. I think it's owing to my
brutal revert of mac_set_unicode_keystrok_event which makes
do_keystrokes dysfunction.
Another possibility is workaround of "export LIBS=-lresolv". I have no
idea. Or owing to merge of multi-tty code?

From reading this group (gmane.emacs.devel) I know Carbon port is
treated as dead (really?). Even though I think I should report my
finding in case someone is willing to try to get it work again.

If Carbon port is treated as dead, and emacs-app (Cocoa port,
http://emacs-app.sf.net) works well, is it possible that emacs-app be
merged (as a branch in cvs or git repo) thus we MacOSX users can have an
evolving Emacs?

Sorry for disturbing everybody.

-- 
Homo sum, humani being nil a me alienum puto

^ permalink raw reply	[flat|nested] 44+ messages in thread