[Biopython-dev] [Bug 2608] Gcc "differ in signedness" warnings with trie.c

Andrew Dalke dalke at dalkescientific.com
Sat Oct 4 22:09:35 UTC 2008


> ------- Comment #1 from biopython-bugzilla at maubp.freeserve.co.uk   
> 2008-10-03 10:35 EST -------
> Interestingly looking at the CVS history, in Bio/triemodule.c  
> revision 1.5 it
> looks like we used to have lots of "char *" casts/variables which  
> were changed
> to "unsigned char *" to solve complaints from the SGI cc compiler  
> (the comment
> doesn't say if these were warnings or errors).

Those were almost certainly warnings.  The "char" type on IRIX is  
unsigned.

I once tracked down a bug in some code which used a char field to store
formal charges.  On IRIX the charges were 0, +1, +2, +254 and +255. :)



				Andrew
				dalke at dalkescientific.com





More information about the Biopython-dev mailing list