[cmucl-help] Re: Building cvs HEAD in Mac OS X (Intel)

2008-05-26 Thread Raymond Toy

Filipe Cabecinhas wrote:


On 25 May, 2008, at 14:24, Raymond Toy wrote:


Filipe Cabecinhas wrote:
Hi. I tried to send this before, but it didn't end up on gmane and I 
wasn't yet subscribed to the list so I didn't see if there were any 
replies.

Yes, cmucl-help became a closed list a few months ago due to spam.
Ah, so that's why I didn't see it. It would be nice if there was a 
warning on cons.org. :-)



I'll try to update the webpages to say so.


Now I have two more questions:
How can I pass compilation flags to the C source-code? I wanted to 
include -I/.../OpenMotif but couldn't and had to resort to macports 
because it's directory was included in the search path.
For motif, you need to go and edit src/motif/server/Config.Darwin.  For 
the other C code, look in src/lisp/Config.x86_darwin.


Any particular reason not to use the macports version?
also: how can I make Emacs show me the source to a CMUCL function When 
I M-. it (Not one of my defined functions but a CMUCL function 
(format, for example))?


When I M-. format and choose #'format, I get:
Error opening #P"target:code/format.lisp", No such file or directory.

If you've installed the code somewhere, there should be a file 
lib/cmucl/lib/site-init.lisp.  In this file, you need to edit the 
search-list target: to point to wherever you stored the sources.   Then 
you should be able to have emacs display the sources.


Ray




[cmucl-help] Re: Building cvs HEAD in Mac OS X (Intel)

2008-05-26 Thread Filipe Cabecinhas

Hi,

Thank you for all the answers.

On 26 May, 2008, at 13:59, Raymond Toy wrote:


For motif, you need to go and edit src/motif/server/Config.Darwin.   
For the other C code, look in src/lisp/Config.x86_darwin.


Any particular reason not to use the macports version?




I just had openMotif installed already. It's no big deal but being  
able to pass compilation options may be nice if I try hacking cmucl ;-)



  - Filipe Cabecinhas






[cmucl-help] Re: Building cvs HEAD in Mac OS X (Intel)

2008-05-26 Thread Filipe Cabecinhas


On 25 May, 2008, at 14:24, Raymond Toy wrote:


Filipe Cabecinhas wrote:
Hi. I tried to send this before, but it didn't end up on gmane and  
I wasn't yet subscribed to the list so I didn't see if there were  
any replies.

Yes, cmucl-help became a closed list a few months ago due to spam.
Ah, so that's why I didn't see it. It would be nice if there was a  
warning on cons.org. :-)



There was an change in the compiler after 19e, so you have to do a  
cross-compile to build it.   Here's how you can do it:


src/tools/create-target.sh 
src/tools/create-target.sh 
src/tools/cross-build-world.sh   src/bootfiles/19e/ 
boot-2008-04-1.lisp <19e-binary>

src/tools/load-world.sh  

Where  and  are different directories, <19e-binary>  
is the path to your 19e binary, and  is the version  
you want to call this.


Then you many want to do a full build using build.sh with this new  
binary.



Nice. Thanks.

Now I have two more questions:
How can I pass compilation flags to the C source-code? I wanted to  
include -I/.../OpenMotif but couldn't and had to resort to macports  
because it's directory was included in the search path.
also: how can I make Emacs show me the source to a CMUCL function When  
I M-. it (Not one of my defined functions but a CMUCL function  
(format, for example))?


When I M-. format and choose #'format, I get:
Error opening #P"target:code/format.lisp", No such file or directory.

Thanks for the reply,

  - Filipe Cabecinhas





[cmucl-help] Re: Building cvs HEAD in Mac OS X (Intel)

2008-05-25 Thread Raymond Toy

Filipe Cabecinhas wrote:
Hi. I tried to send this before, but it didn't end up on gmane and I 
wasn't yet subscribed to the list so I didn't see if there were any 
replies.

Yes, cmucl-help became a closed list a few months ago due to spam.

I'm sorry if it is a duplicate.


I'm trying to build cmucl on my computer but I'm having weird errors.
I downloaded cmucl-19e-x86-darwin and am using it to compile the cvs 
HEAD version, but it throws an error at me while building phase 3.
There was an change in the compiler after 19e, so you have to do a 
cross-compile to build it.   Here's how you can do it:


src/tools/create-target.sh 
src/tools/create-target.sh 
src/tools/cross-build-world.sh   
src/bootfiles/19e/boot-2008-04-1.lisp <19e-binary>

src/tools/load-world.sh  

Where  and  are different directories, <19e-binary> is 
the path to your 19e binary, and  is the version you 
want to call this.


Then you many want to do a full build using build.sh with this new binary.

Let us know how that goes.

Ray







I have the following error in phase 3:
;;; Loading #P"/.../dev/lisp/cmucl/build-2/compiler/x86/vm.x86f".

End-of-File on #"/.../dev/lisp/cmucl/build-2/compiler/x86/vm.x86f">

  [Condition of type END-OF-FILE]

Restarts:
 0: [CONTINUE] Return NIL from load of "vm:vm".
 1:Return NIL from load of "c:loadbackend.lisp".
 2:Return NIL from load of "target:tools/worldload".
 3: [ABORT   ] Return to Top-Level.

Debug  (type H for help)

(LISP::INPUT-UNSIGNED-8BIT-BYTE
#
T
NIL)
Source:
; File: target:code/fd-stream.lisp


And some other errors in vm:*. I also have an error in pcl:
Loading binary of FIXUP...

Unable to display error condition
  [Condition of type UNBOUND-VARIABLE]

Restarts:
 0: [CONTINUE] Return NIL from load of #P"target:pcl/fixup.x86f".
 1:Return NIL from load of "pcl:pclload".

And some others after that. It finally exits with:
gcc -Wstrict-prototypes -Wall -g3 -Di386 -DDARWIN -DGENCGC 
-DLINKAGE_TABLE -iquote . -iquote ../../src/lisp  -Di386 -DDARWIN 
-DGENCGC -DLINKAGE_TABLE  -c -o lisp.o ../../src/lisp/lisp.c

In file included from ../../src/lisp/lisp.c:18:
./internals.h:1:2: error: #error You need to run genesis (via 
build-world.sh) before compiling the startup code!




I noticed that before that, in phase 2, I have (at least) this minor 
message:

;; Loading #P"/.../dev/lisp/cmucl/build-2/assembly/assemfile.x86f".
target:compiler/x86/vm
Error abort.
target:compiler/x86/insts

This is in the logs:
; Compiling DEFINE-STORAGE-CLASSES (CONSTANT CONSTANT):
; Byte Compiling Top-Level Form:
; Compiling DEFINE-STORAGE-CLASSES (CONSTANT CONSTANT):
; Byte Compiling Top-Level Form:

Error in function C::%%DEFCONSTANT:  Constant BYTE-SC-NAMES being 
redefined.


Aborting...
0: (DEBUG:BACKTRACE 536870911
   #)


Can it be failing just due to this error? What can I do to correct it?


Thanks in advance,

 - Filipe Cabecinhas