Re: [OpenBabel-Devel] forcefield classes (isnan)

[Chris M. <c.m...@ga...>]

OBForceField::DistanceGeometry() is still (illegally) setting arrays 
with non-constant dimensions, which won't compile for me (Visual C++8).

It is the #define isnan code in forcefied.h that prevents compilation 
for some reason, and needs to be deleted.

It looks like the compiler you are using is a bit too forgiving. (Mine, 
also - last week it didn't catch #elif instead of #else.)

Chris

Tim Vandermeersch wrote:
> Other people were having the same problem. The latest focefield.cpp
> (just commited to SVN) uses OBIsNan(). This is defined in the file
> itself. (could be moved to include/openbabel/math/?? to be used by
> other files)
> 
> Tim
> 
> On 28 Feb 2007 17:13:02 +0000, N.W. England <nw...@ca...> wrote:
>> I was trying to compile the latest svn version(revision 1807) of openbabel
>> under SUSE with gcc (version 3.3.5 20050117 (prerelease) (SUSE Linux)) and
>> it was still failing due to forcefield.cpp line 153 complaining about
>> invalid type of 'isnan'(and many other lines) I found this odd as the macro
>> on that line doesn't seem to expand into anything with an isnan in it.
>>
>> After installing the latest version of gcc(version 4.1.1) it compiles
>> perfectly and passes all unit tests. (after installing libxml-devel)
>>
>> I still have the old version of gcc if further testing is needed.
>>
>> - Nick
>>
>> On Feb 20 2007, Geoffrey Hutchison wrote:
>>
>>> On Feb 20, 2007, at 12:46 PM, Tim Vandermeersch wrote:
>>>
>>>> Line 401 was taken from the documentation api so I expected it to
>>>> work. But std::vector may indeed be more portable...
>>> Ah, but that was documentation I wrote using GCC. I'll go fix that.
>>>
>>> Cheers,
>>> -Geoff
>