[SciPy-user] ***[Possible UCE]*** Re: Failed Build and Multiplication Question

[Travis Oliphant]

David M. Cooke wrote:

>On Jan 12, 2006, at 21:50 , Travis Oliphant wrote:
>
>  
>
>>Travis Brady wrote:
>>
>>    
>>
>>>Here is all of the output:
>>>
>>>      
>>>
>>This is fixed in SVN.   There was a mistake in when the expm1f,  
>>expm1l,
>>logp1f, and logp1l functions should be defined...
>>    
>>
>
>I had added log1p and exp1m to the same area that handles atanf (and  
>others) not existing, so they were.
>  
>
Yes, I saw that,  but the problem was that on my  windows platform 
(compiled using mingw32)  the config.h showed HAVE_LOG1P  but no 
HAVE_EXPM1 but also HAVE_LONGDOUBLE_FUNCS and HAVE_FLOAT_FUNCS were 
defined.

So, my system did not have expm1f nor expm1l even though it had expf and 
expl.  Thus, the single HAVE_FLOAT_FUNCS check wasn't enough for these 
extra C99 functions. 

>The error from his compile indicates that he has expm1f and expm1l,  
>it's just that they're done using a #define instead of functions (so  
>the address can't be taken of them).
>  
>
I guess I didn't look closely,  I just got a similar error regarding 
exmp1f and exm1l not being defined and started trying to fix it.

Thanks for looking at this,

-Travis