[CM] clm-2 OS X woes

Michael Klingbeil michael@klingbeil.com
Wed, 25 Aug 2004 03:40:19 -0400

Thanks for the config suggestion. In fact there were a bunch of other 
configuration problems that needed to be sorted out. The last version 
of clm-2 (June 30, 2004) expected that the working directory was set 
to the clm-2 directory so config.h was never getting built and then 
it was dumped into the wrong place.

Also *cflags* in all.lisp needed " -I" clm-directory and similar 
changes were needed in defins.lisp.

I noticed that most of these things made their way into clm-3 [ 
although I think line 160 in defins.lisp for clm-3 5-Aug still needs 
the following:

(defvar *cflags* (concatenate 'string " -I" clm-directory
	" -O2 -g"  ; the -O2 is needed!  strange errors from ACL if 
it's omitted


In any case since things have branched to clm-3 perhaps these clm-2 
fixes are less relevant. I thought I would send this to the list 
anyway in case people are still doing clm-2 stuff.

>  > _ccos
>>  ...
>>  Any suggestions before I start ripping things apart? Should I be 
>>using clm-3?
>Yes to the latter, but if you want to use CLM-2, the error
>has to do with the HAVE_COMPLEX_TRIG switch -- set it to
>0 or undefined and the code that depends on it should be
>(ccos is "complex cosine" -- it's in current versions of
>gcc as part of the built-in, but not well publicized complex
>number support.  It's needed by the Dolph-Chebychev window
>because the algorithm sometimes needs values such as the
>arc-cosine of a number greater than 1.0 -- i.e. it has
>wandered off the real line and into the trackless wastes
>of the complex plane.)