Emacs CVS on OSX Tiger (10.4)

Emacs CVS on OSX Tiger (10.4)

[Andrew Beekhof]

Hi All,

It appears that 10.4 has changed enough compared to 10.3.x to break 
Emacs.

Pre-compiled packages built for 10.3.x break during startup, so I tried 
compiling from source.

I ran into a couple of problems which I was able to work around, but 
without really knowing what I was doing, probably did so in the worst 
way possible.

Problem #1
macgui.c was trying to use emacs_mktime and failed during the link 
stage.
	I hunted the entire codebase and couldn't find this 
function/macro/other anywhere... am I missing something?
	In the end i commented out these lines:
./src/macgui.h-46-#undef mktime
./src/macgui.h:47:#define mktime emacs_mktime

Problem #2
sys/time.h (for reasons that are beyond me) now defines select() and a 
number of other bits and pieces that are also in sys/types.h.  Maybe 
thats not the whole cause, but to make a long story short, mac.c 
(specifically sys_select() and select_and_poll_event() ) doesn't 
compile.
	So I removed both functions and, well, it seems to work anyway....
	Here are the compiler errors:
<snip>
mac.c: In function 'select_and_poll_event':
mac.c:3877: error: request for member 'fds_bits' in something not a 
structure or union
mac.c: In function 'sys_select':
mac.c:3895: error: argument 'rfds' doesn't match prototype
/usr/include/sys/time.h:242: error: prototype declaration
mac.c:3895: error: argument 'wfds' doesn't match prototype
/usr/include/sys/time.h:242: error: prototype declaration
mac.c:3895: error: argument 'efds' doesn't match prototype
/usr/include/sys/time.h:242: error: prototype declaration
mac.c:3901: error: request for member 'fds_bits' in something not a 
structure or union
mac.c:3904: error: request for member 'fds_bits' in something not a 
structure or union
mac.c:3918: error: request for member 'fds_bits' in something not a 
structure or union
mac.c:3929: error: request for member 'fds_bits' in something not a 
structure or union
mac.c: In function 'mac_get_system_locale':
mac.c:4224: warning: pointer targets in passing argument 4 of 
'LocaleRefGetPartString' differ in signedness
mac.c:4226: warning: pointer targets in passing argument 1 of 
'build_string' differ in signedness
make: *** [mac.o] Error 1


If anyone has suggestions on how I should have fixed the above, I can 
have a go and report back with a patch.

Andrew

--
Andrew Beekhof

"If it weren't for my horse, I wouldn't have spent that year in 
college" - Unknown, courtesy of Lewis Black

Re: Emacs CVS on OSX Tiger (10.4)

[John Owens]

Andrew Beekhof <beekhof <at> gmail.com> writes:

> 
> Hi All,
> 
> It appears that 10.4 has changed enough compared to 10.3.x to break 
> Emacs.

I can confirm this is a problem and wanted to detail my similar workaround.

1) Comment out select_and_poll_event and sys_select function def'ns in mac.c.
2) Comment out the mktime -> emacs_mktime definition for Carbon in macgui.h.
3) In macterm.c and process.c, #undef select (after include files)
4) For good measure, in sysselect.h, comment out #define select sys_select

I'm certainly not sure what the proper way to handle the build problems, but 
at least this gives me a usable emacs for the time being.

JDO

Re: Emacs CVS on OSX Tiger (10.4)

[Andrew Beekhof]

On Apr 22, 2005, at 10:36 PM, John Owens wrote:

> Andrew Beekhof <beekhof <at> gmail.com> writes:
>
>>
>> Hi All,
>>
>> It appears that 10.4 has changed enough compared to 10.3.x to break
>> Emacs.
>
> I can confirm this is a problem and wanted to detail my similar 
> workaround.
>
> 1) Comment out select_and_poll_event and sys_select function def'ns in 
> mac.c.
> 2) Comment out the mktime -> emacs_mktime definition for Carbon in 
> macgui.h.
> 3) In macterm.c and process.c, #undef select (after include files)
> 4) For good measure, in sysselect.h, comment out #define select 
> sys_select
>
> I'm certainly not sure what the proper way to handle the build 
> problems, but
> at least this gives me a usable emacs for the time being.
>

Yeah sorry about that, I neglected to mention 3 & 4.

The macgui.h fix seems obvious and probably safe, but I'm unsure about 
the sys_select since previous commit messages specifically indicate 
that it should be used.

I too am just happy I can use my favorite editor again (wasted a whole 
weekend trying to something that came close), but if someone indicates 
a preferred direction I'm also willing to create and submit a patch 
too.

> JDO
>
>
>
>
> _______________________________________________
> Emacs-devel mailing list
> Emacs-devel@gnu.org
> http://lists.gnu.org/mailman/listinfo/emacs-devel
>
>
--
Andrew Beekhof

"Ooo Ahhh, Glenn McRath" - TISM