[chuck-users] adsr

[Spencer Salazar]

Eduard/Jukka,
Could you edit line 7 your makefile like to look like this:
FLAGS=-D__MACOSX_CORE__ -D__LITTLE_ENDIAN__ -c

and try recompiling?

On our PPC systems, __LITTLE_ENDIAN__ is predefined when cross- 
compiling for x86, and I had thought that that would be the case on  
x86 systems also.  However, judging by the nature of your build  
errors, it seems that this preprocessor macro might not be present.

spencer

On Sep 11, 2006, at 7:20 PM, Jukka Akkanen wrote:

>
> On Sep 11, 2006, at 5:06 PM, Ge Wang wrote:
>
>> Hi Eduard!
>>
>>> When building chuck from cvs in macintel got the following:
>>>
>>> gcc -D__MACOSX_CORE__ -O3 -c chuck_instr.cpp
>>> /usr/include/sys/types.h:86: error: declaration does not declare
>>> anything
>>
>> This is quite strange.  We don't have a macintel around, so we'll try
>> to debug remotely.  I checked in a slight change to the offending
>> file,
>> please update and try compiling again?
>>
>> Also:
>> 1. what gcc version on macintel?
>> 2. does the release 1.2.0.6 compile or do you get the same error?
>>
>> Anyone else encounter the same problem?
>
> I am seeing this, too. First, a disclaimer. I've barely gotten into
> using ChucK and hadn't looked at source until I heard about this
> problem. Somehow it wasn't very easy for me to find information about
> how to check out the source from CVS, but I found it in the chuck-dev
> mailing list archive, eventually. Maybe it's time I subscribe there,
> too.
>
> The problem is that chuck_instr.h has this:
>
>      #define uint unsigned long
>
> and sys/types.h has:
>
>      #ifndef _POSIX_C_SOURCE
>      ...
>      typedef unsigned int            uint;           /* Sys V
> compatibility */
>      #endif
>
> which results in a:
>
>      typedef unsigned int unsigned long;
>
> Fixing this is not the only hurdle, later on you'll see things like:
>
>      /var/tmp//ccAPkhiP.s:20471:no such 386 instruction: `fctiw'
>      /var/tmp//ccAPkhiP.s:20472:no such 386 instruction: `stfd'
>
> which is caused by inline assembly for the wrong architecure getting
> picked up from util_sndfile.h (looks like OS X does have lrint() and
> lrintf() so there should be no need for asm at all).
>
> An Intel Mac shouldn't be necessary to debug these kinds of issues.
> All you need is recent developer tools with universal libraries and
> you'll be able to specify "-arch i386" in compiler and linker flags
> to produce Intel binaries on a PPC machine.
>
> -Jukka
> _______________________________________________
> chuck-users mailing list
> chuck-users at lists.cs.princeton.edu
> https://lists.cs.princeton.edu/mailman/listinfo/chuck-users