Re: 0.1.14 is out/0.1.15 needed

Wolfgang Mueller wrote:

>Dear David,
>
>On Thu, 31 Mar 2005, David Squire wrote:
>
>
>  
>
>>It appears that there was something of a communication break-down on the 
>>changes needed. There are no 64 bit problems with feature extraction 
>>that I am aware of, but some header issues that appear to affect 
>>different linux distributions differently.
>>    
>>
>
>I was very much aware of problems with 0.1.14 today. They call it "segmentation fault". I think it is the header you spoke of. I guess we should be able to sort this out quickly. 
>  
>
Hi all,

I've been doing some testing, and the suspicions above seem to be 
confirmed.  0.1.14 from the ftp site installs and runs perfectly on my 
32 bit Pentium III RedHat 9 system. I tested adding a new collection, 
and querying the result, as well as querying collections added by other 
versions. To get it working on my AMD 64 Debian system, I need to fix 
the malloc.h issue described in 
http://lists.gnu.org/archive/html/bug-gift/2005-03/msg00001.html. This 
supports my suspicion that the malloc.h thing is a linux distro or gcc 
version issue, rather than a 32/64 bit thing.

In short, if you want to get 0.1.14 working while you wait for 0.1.15, I 
suggest that you add the line

#include <malloc.h>

after

#include <math.h>

at the tops of rgb2hsv_utils.c and quantize.c in the directory 
FeatureExtraction/, and then recompile and install.

If this doesn't fix the feature extraction problem, please let us know 
ASAP - that would mean that there is another problem of which I am unaware.

Regards,

David

-- 
Dr. David McG. Squire,  Associate Head, Caulfield Campus
Faculty of Information Technology, Monash University, Australia
CRICOS Provider No. 00008C       http://www.csse.monash.edu.au/~davids/