[OSX] G4 compiled binary fails on G5

[OSX] G4 compiled binary fails on G5

[Sébastien Kirche]

Hi,

some days  ago I  published a  pre-compiled OSX binary  built from  the CVS.
Craig  A. Finseth  told me  that  this version  fails during  launch on  his
system.

Looking at the crash log in the console utility shown the following :

	dyld: /Applications/Emacs.app/Contents/MacOS/Emacs version
	mismatch for library: /usr/lib/libncurses.5.dylib (compatibility
	version of user: 6.0.0 greater than library's version: 5.0.0)

I built Emacs on my G4/10.3.7, and Craig has a G5/10.3.7.

Actually that message does not tell me if the problem is on my part. I don't
understand the  last sentence  : it seems  that I  have the 5.0  version and
Craig the  6.0 but  in that  case i don't  understand why  it fails.  If the
application is  linked against  v.5 library, it  should work with  a greater
version, shouldn't it ?

Could an OSX familiar give me some clue ? Thank you.

-- 
Sébastien Kirche

Re: [OSX] G4 compiled binary fails on G5

[Steven Tamm]

I'm running 10.3.7 on a G5 and only have ncurses.5.  However I have had 
this problem before.

My suspicion is that you have fink installed on your machine and the 
other person does not.  For example, I have fink and my build uses 
/sw/lib/ncurses.  Fink's ncurses is incompatible with the normal one by 
design.  If you look in the Makefile, the CC line will be "gcc 
-I/sw/include -L/sw/liib"  This is very useful for most builds, but not 
for binary builds.

You should recompile and either run
make CC=gcc
or
make-package -M,CC=gcc

-Steven

On Jan 31, 2005, at 2:15 AM, Sébastien Kirche wrote:

> Hi,
>
> some days  ago I  published a  pre-compiled OSX binary  built from  
> the CVS.
> Craig  A. Finseth  told me  that  this version  fails during  launch 
> on  his
> system.
>
> Looking at the crash log in the console utility shown the following :
>
> 	dyld: /Applications/Emacs.app/Contents/MacOS/Emacs version
> 	mismatch for library: /usr/lib/libncurses.5.dylib (compatibility
> 	version of user: 6.0.0 greater than library's version: 5.0.0)
>
> I built Emacs on my G4/10.3.7, and Craig has a G5/10.3.7.
>
> Actually that message does not tell me if the problem is on my part. I 
> don't
> understand the  last sentence  : it seems  that I  have the 5.0  
> version and
> Craig the  6.0 but  in that  case i don't  understand why  it fails.  
> If the
> application is  linked against  v.5 library, it  should work with  a 
> greater
> version, shouldn't it ?
>
> Could an OSX familiar give me some clue ? Thank you.
>
> -- 
> Sébastien Kirche
>
>
>
> _______________________________________________
> Emacs-devel mailing list
> Emacs-devel@gnu.org
> http://lists.gnu.org/mailman/listinfo/emacs-devel

Re: [OSX] G4 compiled binary fails on G5

[Sébastien Kirche]

Le 31 Jan 2005, Steven Tamm a formulé :

> I'm running 10.3.7 on a G5 and only have ncurses.5.  However I have
> had this problem before.
> 
> My suspicion is that you have fink installed on your machine and the
> other person does not. 

I confirm that i do have Fink  installed. I dunno for the machine that shown
the problem but it is possible.

> For example, I have fink and my build uses /sw/lib/ncurses. Fink's ncurses
> is  incompatible  with the  normal  one  by design.  If  you  look in  the
> Makefile, the CC line will  be "gcc -I/sw/include -L/sw/liib" This is very
> useful for most builds, but not for binary builds.
> 
> You should recompile and either run
> make CC=gcc
> or
> make-package -M,CC=gcc

Thank you for that explanation, and for the fix tip. 

I will try another build ASAP.

-- 
Sébastien Kirche

Re: [OSX] G4 compiled binary fails on G5

[Kim F. Storm]

Steven Tamm <steventamm@mac.com> writes:

> I'm running 10.3.7 on a G5 and only have ncurses.5.  However I have
> had this problem before.
>
> My suspicion is that you have fink installed on your machine and the
> other person does not.  For example, I have fink and my build uses
> /sw/lib/ncurses.  Fink's ncurses is incompatible with the normal one
> by design.  If you look in the Makefile, the CC line will be "gcc
> -I/sw/include -L/sw/liib"  This is very useful for most builds, but
> not for binary builds.
>
> You should recompile and either run
> make CC=gcc
> or
> make-package -M,CC=gcc

Maybe this should be added to etc/PROBLEMS.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: [OSX] G4 compiled binary fails on G5

[Sébastien Kirche]

Le 31 Jan 2005, Steven Tamm vraute :

> You should recompile and either run
> make CC=gcc
> or
> make-package -M,CC=gcc

I have compiled a  version with «make-package --self-compiled -M,CC=gcc» and
now wait for feedback.

I also made an OS9 version with the USE_LSB_TAG and i wonder if it should be
given also to OSX ?

That symbol seems to be defined on  mac if __MRC__ is defined. IIRC MrC used
to be the OS9 compiler from Apple/MPW.

If so,  looking dumb  I did not  figured the  syntax to provide  both CC=gcc
*and* USE_LSB_TAG or maybe -DUSE_LSB_TAG...

Can you tell me the correct syntax to define some symbols from command line
without modifying the source and specifying the cc or cflags ?

Thank you.

-- 
Sébastien Kirche

Re: [OSX] G4 compiled binary fails on G5

[Stefan Monnier]

> > You should recompile and either run
> > make CC=gcc
> > or
> > make-package -M,CC=gcc
> 
> I have compiled a  version with «make-package --self-compiled -M,CC=gcc» and
> now wait for feedback.
> 
> I also made an OS9 version with the USE_LSB_TAG and i wonder if it should be
> given also to OSX ?

How did you set the USE_LSB_TAG ?  What compiler did you use?

> That symbol seems to be defined on  mac if __MRC__ is defined. IIRC MrC used
> to be the OS9 compiler from Apple/MPW.

I must say I have no clue under which circumstance are the mac/inc/*
files used.  Is that for Mac OS 9 only?

While I'm looking at them, couls someone tell me what the following comment
(in mac/inc/config.h) refers to:

   /* Use low-bits for tags.  If ENABLE_CHECKING is turned on together
      with USE_LSB_TAG, optimization flags should be explicitly turned
      off.  */

what's the problem with optimization flags and ENABLE_CHECKING and
USE_LSB_TAG?

Also in s-mac.h, would it be correct to add to the code

   #ifdef __MRC__
   #define DECL_ALIGN(type, var) type var
   #endif

a little comment saying something like "MrC always aligns structures larger
than 8bytes on 8byte boundaries".  If not, what is the justification for
defining DECL_ALIGN as a nop?

> If so,  looking dumb  I did not  figured the  syntax to provide  both CC=gcc
> *and* USE_LSB_TAG or maybe -DUSE_LSB_TAG...

If you tell me how you set USE_LSB_TAG, I may be able to tell you how to
combine the two.


        Stefan

Re: [OSX] G4 compiled binary fails on G5

[Steven Tamm]

On Feb 2, 2005, at 9:21 AM, Stefan Monnier wrote:

>>> You should recompile and either run
>>> make CC=gcc
>>> or
>>> make-package -M,CC=gcc
>>
>> I have compiled a  version with «make-package --self-compiled 
>> -M,CC=gcc» and
>> now wait for feedback.
>>
>> I also made an OS9 version with the USE_LSB_TAG and i wonder if it 
>> should be
>> given also to OSX ?
>
> How did you set the USE_LSB_TAG ?  What compiler did you use?

USE_LSB_TAG is defined for MAC_OSX.  (see lisp.h)

>
>> That symbol seems to be defined on  mac if __MRC__ is defined. IIRC 
>> MrC used
>> to be the OS9 compiler from Apple/MPW.
>
> I must say I have no clue under which circumstance are the mac/inc/*
> files used.  Is that for Mac OS 9 only?

Yes.  Only for OS9.  The OSX stuff doesn't use anything in the mac 
directory except for Emacs.app, which houses the resource files for 
generating the application.  These probably should be autogenerated at 
some point.

-Steven

Re: [OSX] G4 compiled binary fails on G5

[Stefan Monnier]

>>> I also made an OS9 version with the USE_LSB_TAG and i wonder if it
>>> should be given also to OSX ?
>> 
>> How did you set the USE_LSB_TAG ?  What compiler did you use?

> USE_LSB_TAG is defined for MAC_OSX.  (see lisp.h)

Yes, my question is about his compiling the MacOS9 version with USE_LSB_TAG
(I know Mac OS X uses USE_LSB_TAG, since I put that part of the code in
lisp.h myself ;-)


        Stefan

Re: [OSX] G4 compiled binary fails on G5

[Stefan Monnier]

> Now, i just  checked in the compilation script that  the compiled invoked is
> actually MrC. Thus  i suppose that during compilation  it defines __MRC__ so
> my  modification  is  useless and  the  limitation  of  512 MB  was  already
> removed... and my question is non sense.

Thank you for checking it.  So we don't need to update the config
files.  Good.

> Is there a method to test for that limitation on a 512 MB machine ? Is there
> any  mean to  know  by asking  Emacs  what symbols/options  were defined  at
> compile time ?

In general, there isn't but in this specific case there is: attach your
Emacs process to a debugger and check the value of the variable
`gdb_use_lsb'.

It's actually used in emacs/src/.gdbinit so that the gdb macros that analyse
Lisp_Object values can do it correctly in the various different
tagging cases.


        Stefan

Re: [OSX] G4 compiled binary fails on G5

[Stefan Monnier]

> Does it mean that OSX is not affected at all by any memory limitation ?

Yup,


        Stefan

Mac OS 9 support

[YAMAMOTO Mitsuharu]

[-- Attachment #1: Type: text/plain, Size: 2918 bytes --]

I'm not sure how many people are interested in Mac OS 9 support, but
now it works again.  Although the source codes for Mac OS 9 have been
bundled with the standard distribution, it could not be compiled as of
Emacs 21.3.

Changes (from 21.2):
  - Works with >256MB memory systems.
  - Most of display features such as images, toolbars, and tooltips
    are available.

Problems:
  - No support for Metrowerks CodeWarrior Pro 6.  Only tested with
    MPW-GM, which is available free of charge.
  - I'm not sure how old versions of Mac OS 8/9 are compatible with.
    I tested it with Mac OS 9.2.2/QuickTime 6.0.3/CarbonLib 1.6.1
    running on the Classic environment in Mac OS X 10.3.5.
  - USE_LSB_TAG is defined so that it can work with >256MB memory
    system.  It seems to work with DECL_ALIGN as the identity macro,
    but not 100% sure.
  - "#define GC_MARK_STACK GC_MAKE_GCPROS_NOOPS" also seems to work,
    but not defined in the attached patch.
  - It's almost impossible to select an item from a pop-up menu
    invoked from a toolbar button, because the mouse button is already
    released when the menu appears.

Install:

  The first half of the steps requires Mac OS X and its Developer Tools.

  1. Get the Emacs source as usual.

    OSX% mkdir somedir
    OSX% cd somedir
    OSX% env CVS_RSH=ssh cvs -z3 -d:ext:anoncvs@savannah.gnu.org:/cvsroot/emacs co emacs

  2. Apply the attached patch.

    OSX% cd emacs
    OSX% zcat someplace/diff-classic.gz | patch -p0

  3. Setup resource forks.

    OSX% /Developer/Tools/SetFile -t TEXT mac/inc/* mac/inc/*/* mac/src/* lib-src/* src/*

  4. Generate .elc files.

    OSX% mkdir ../emacs-carbon
    OSX% cd ../emacs-carbon
    OSX% ../emacs/configure --without-x
    OSX% make bootstrap

  You can do the remaining steps on "real" Mac OS 9, which is much
  faster than the Classic environment in Mac OS X.

  5. Download MPW-GM from Apple.

    http://developer.apple.com/tools/mpw-tools/
    ftp://ftp.apple.com/developer/Tool_Chest/Core_Mac_OS_Tools/MPW_etc./MPW-GM_Images/MPW-GM.img.bin

  6. Extract and install it into an appropriate place.
     (e.g. "/Applications (Mac OS 9)").

  7. Invode MPW Shell.

    OSX% open "/Applications (Mac OS 9)/MPW-GM/MPW/MPW Shell"

  8. Do the following inside MPW Shell.  Note that no prompt such as
     "MPW>" is displayed in reality, and you must hit "enter" instead of
     "return" to execute each command.

    MPW> Directory "Macintosh HD:Users:yourname:somedir:emacs:mac"
    MPW> make Emacs -f makefile.MPW > Emacs.MakeScript
    MPW> Emacs.MakeScript

  9. Wait for some time until the left upper corner's message becomes
     "MPW Shell".  It will take much time to execute "make-docfile"
     under the Classic environment even on Power Mac G5.

  10. Invoke "Emacs MPW" (white bear icon) in emacs/mac directory.

  11. Have fun!!

				     YAMAMOTO Mitsuharu
				mituharu@math.s.chiba-u.ac.jp

[-- Attachment #2: ChangeLog.gz --]
[-- Type: application/octet-stream, Size: 1445 bytes --]

[-- Attachment #3: diff-classic.gz --]
[-- Type: application/octet-stream, Size: 9246 bytes --]

[-- Attachment #4: Type: text/plain, Size: 142 bytes --]

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

Re: Mac OS 9 support

[Stefan Monnier]

>   - USE_LSB_TAG is defined so that it can work with >256MB memory
>     system.  It seems to work with DECL_ALIGN as the identity macro,
>     but not 100% sure.

If you want to be more confident that it does indeed work, compile
with -DENABLE_CHECKING.  This should pretty quickly catch any problem.
I think it's not unexpected on RISC systems like PowerPC that DECL_ALIGN is
not needed.

>   - "#define GC_MARK_STACK GC_MAKE_GCPROS_NOOPS" also seems to work,
>     but not defined in the attached patch.

It should work, and it's probably better (i.e. more robust) to define it.


        Stefan

Re: Mac OS 9 support

[YAMAMOTO Mitsuharu]

>>>>> On Sat, 16 Oct 2004 14:37:39 -0400, Stefan Monnier <monnier@iro.umontreal.ca> said:

> If you want to be more confident that it does indeed work, compile
> with -DENABLE_CHECKING.  This should pretty quickly catch any
> problem.  I think it's not unexpected on RISC systems like PowerPC
> that DECL_ALIGN is not needed.

Actually, I used this in the early development stage.  It seems that
the default optimization option of MrC compiler is not compatible with
-DENABLE_CHECKING.  So those who want to use this should turn off the
optimization flags as I wrote in the patch for mac/inc/s-mac.h.

  MPW> make Emacs -f makefile.MPW -d OptOption="-opt off" > Emacs.MakeScript

>> - "#define GC_MARK_STACK GC_MAKE_GCPROS_NOOPS" also seems to work,
>> but not defined in the attached patch.

> It should work, and it's probably better (i.e. more robust) to
> define it.

Thanks for the comments.  As far as I tested, GC_MAKE_GCPROS_NOOPS
seems to work well with the default optimization option.  But it
crashes at the startup if used with "-opt off" option above for
unknown reasons.

				     YAMAMOTO Mitsuharu
				mituharu@math.s.chiba-u.ac.jp

Re: [OSX] G4 compiled binary fails on G5

[YAMAMOTO Mitsuharu]

>>>>> On Wed, 02 Feb 2005 12:21:34 -0500, Stefan Monnier <monnier@iro.umontreal.ca> said:

> what's the problem with optimization flags and ENABLE_CHECKING and
> USE_LSB_TAG?

It crashes at startup time (while loading loadup.el).  This issue was
mentioned in my previous mail.

>>>>> On Sun, 17 Oct 2004 18:23:32 +0900, YAMAMOTO Mitsuharu <mituharu@math.s.chiba-u.ac.jp> said:

>> If you want to be more confident that it does indeed work, compile
>> with -DENABLE_CHECKING.  This should pretty quickly catch any
>> problem.  I think it's not unexpected on RISC systems like PowerPC
>> that DECL_ALIGN is not needed.

> Actually, I used this in the early development stage.  It seems that
> the default optimization option of MrC compiler is not compatible
> with -DENABLE_CHECKING.

Unfortunately, I still don't know why, and it's difficult for me to
use a debugger on Mac OS 9.  I couldn't invoke PowerMac Debugger
(http://developer.apple.com/tools/debuggers/) on the Classic
Environment.

>>>>> On Wed, 02 Feb 2005 12:21:34 -0500, Stefan Monnier <monnier@iro.umontreal.ca> said:

> Also in s-mac.h, would it be correct to add to the code

>    #ifdef __MRC__
>    #define DECL_ALIGN(type, var) type var
>    #endif

> a little comment saying something like "MrC always aligns structures
> larger than 8bytes on 8byte boundaries".  If not, what is the
> justification for defining DECL_ALIGN as a nop?

Sorry, I'm not 100% sure.  All I have is "Just it seems to work" and a
description in "Macintosh C/C++ ABI":
(http://developer.apple.com/tools/mpw-tools/compilers/docs/abi_spec.pdf)

  3.3.1.1 Natural Alignment

  The natural alignment of the type of is used whenever an instance of
  that type, a local or global, is allocated to memory or assigned a
  memory address.

				 ...

  The natural alignment for an aggregate type is the maximum of the
  natural alignments of its members.

According to the natural alignment, one can only assume that Lisp_Subr
is aligned on 4 byte boundaries.  But there's also the following note:

  NOTE: It is recommended that compilers use the following formula to
  choose the alignment for local and global aggregate variables
  instead of using the natural alignment:

				 ...

     size of 8-15 bytes -> alignment of 8 bytes
     size of 16 byte or more -> larger of 8 and the embedding
                                alignment of the variable's type
     
So, 8 byte alignment would be expected.  But again, I'm not 100% sure.

				     YAMAMOTO Mitsuharu
				mituharu@math.s.chiba-u.ac.jp

Re: Mac OS 9 support (and some enhancements for Carbon Emacs)

[YAMAMOTO Mitsuharu]

[-- Attachment #1: Type: text/plain, Size: 1540 bytes --]

>>>>> On Sat, 16 Oct 2004 16:28:55 +0900, YAMAMOTO Mitsuharu <mituharu@math.s.chiba-u.ac.jp> said:

> I'm not sure how many people are interested in Mac OS 9 support, but
> now it works again.

A revised Mac OS 9 support patch is attached.  It also contains the
following enhancements for Carbon Emacs on Mac OS X.

  * Rearrangement of startup routine
    - The Carbon build no longer crashes with window system specific
      functions such as `xw-display-color-p' when invoked with `-nw'
      option.
    - Many window system command-line options are now interpreted.
  * WM size hints emulation
    - If the geometry is not specified explicitly, the initial frame
      is positioned at the center of screen.  A new frame is placed
      near the selected frame (10.2 and later).
    - The frame (border) that is displayed during window resize now
      changes its size according to the character boundary.
  * `select' emulation with CFSocket (not enabled by default)
    - It reduces CPU usage when Emacs is idle even with asynchronous
      processes.  Previously, polling with a short interval has been
      used for simultaneously monitoring window events and process
      outputs.  Although the new one is more efficient and works fine
      as long as I tested, it is not enabled by default in case it may
      cause unpredictable results because it spawns a thread for
      monitoring sockets.  To enable it, specify
      `-DSELECT_USE_CFSOCKET' for CFLAGS.

				     YAMAMOTO Mitsuharu
				mituharu@math.s.chiba-u.ac.jp

[-- Attachment #2: ChangeLog.gz --]
[-- Type: application/octet-stream, Size: 2820 bytes --]

[-- Attachment #3: diff-classic-etc.gz --]
[-- Type: application/octet-stream, Size: 37452 bytes --]

[-- Attachment #4: Type: text/plain, Size: 142 bytes --]

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

Re: Mac OS 9 support (and some enhancements for Carbon Emacs)

[Steven Tamm]

All of these changes have been checked in to latest CVS.  Make sure you  
recompile the lisp directory (or remove the term/mac-win.elc file).   
Otherwise you may get a error saying "Mac native windows not in use" or  
"Mac native windows are not in use".

I checked in the MacOS 9 portions without testing them personally: I  
would ask that people that are still using the classic version try it  
out.  As for the OSX changes, the change to the resizing behavior is  
quite awesome.

As for the CFSocket stuff, I tried it with CFLAGS="-g -O2  
-DSELECT_USE_CFSOCKET" and tried my standard battery of tests.  It  
seems that it's working just fine.  I would ask all OSX users that do a  
lot of subprocess work (gnus, tex, etc.) to try it with it turned on  
and report if there are any hangs or weird behaviors.

Thanks,
-Steven

On Dec 24, 2004, at 12:52 AM, YAMAMOTO Mitsuharu wrote:

>>>>>> On Sat, 16 Oct 2004 16:28:55 +0900, YAMAMOTO Mitsuharu  
>>>>>> <mituharu@math.s.chiba-u.ac.jp> said:
>
>> I'm not sure how many people are interested in Mac OS 9 support, but
>> now it works again.
>
> A revised Mac OS 9 support patch is attached.  It also contains the
> following enhancements for Carbon Emacs on Mac OS X.
>
>   * Rearrangement of startup routine
>     - The Carbon build no longer crashes with window system specific
>       functions such as `xw-display-color-p' when invoked with `-nw'
>       option.
>     - Many window system command-line options are now interpreted.
>   * WM size hints emulation
>     - If the geometry is not specified explicitly, the initial frame
>       is positioned at the center of screen.  A new frame is placed
>       near the selected frame (10.2 and later).
>     - The frame (border) that is displayed during window resize now
>       changes its size according to the character boundary.
>   * `select' emulation with CFSocket (not enabled by default)
>     - It reduces CPU usage when Emacs is idle even with asynchronous
>       processes.  Previously, polling with a short interval has been
>       used for simultaneously monitoring window events and process
>       outputs.  Although the new one is more efficient and works fine
>       as long as I tested, it is not enabled by default in case it may
>       cause unpredictable results because it spawns a thread for
>       monitoring sockets.  To enable it, specify
>       `-DSELECT_USE_CFSOCKET' for CFLAGS.
>
> 				     YAMAMOTO Mitsuharu
> 				mituharu@math.s.chiba-u.ac.jp
> <ChangeLog.gz><diff-classic- 
> etc.gz>_______________________________________________
> Emacs-devel mailing list
> Emacs-devel@gnu.org
> http://lists.gnu.org/mailman/listinfo/emacs-devel

Re: Mac OS 9 support (and some enhancements for Carbon Emacs)

[Sébastien Kirche]

Le 24 Dec 2004, YAMAMOTO Mitsuharu s'est exprimé ainsi :

> > I'm not sure how many people are interested in Mac OS 9 support, but
> > now it works again.
> 
> A revised Mac OS 9 support patch is attached.  It also contains the
> following enhancements for Carbon Emacs on Mac OS X.

Cheers Mitsuharu !!!
I have at last managed to build a fresh 21.3.50 OS9 version. Thanks !

Now, i suppose that  if i need OS9 Emacs, maybe there  are other people that
need it too so i would like to host that version.

Could Emacs folks give me some directives for that purpose ?
- is there any objection
- how should  i package it  ? i think  of a mac  .sit archive for  the whole
  emacs directory

What do you think ?

-- 
Sébastien Kirche