Re: gift-write-feature-descs segfaults... (is the GIFT broken?)

Jonas:

This is undoubtedly my fault, for not checking the other binaries.

I've been under the weather recently, and i've only been checking
the gift-extract-features binary.

I'll try to get a fix in sometime this evening..

Yes, normally C screams and yells at you, but the gift was written
without function prototypes, so the compiler cant tell, until link time.

I'll also whip up a quick patch adding function prototypes, and post
it to the mailing list, for review.

Back, after being food poisoned,

Julia Longtin <[email protected]>

On Sun, Aug 27, 2006 at 04:02:46PM +0200, Jonas Lindqvist wrote:
> Hi!
> 
> I'm (attempting to be) using the latest CVS version.
> 
> gift-write-feature-descs reads a PPM file on STDIN, and is supposed  
> to output some kind of data that can be redirected into  
> InvertedFileFeatureDescription.db.
> 
> However, now it crashes with a segmentation fault.
> 
> Here's what happens:
> The file write_feature_descs.c, uses a function called  
> init_feature_variables. It uses it with an int (colmap_size) as its  
> only parameter.
> So where is this function defined? There is in fact a similar  
> function in extract_block_features.c, but it takes no less than FOUR  
> parameters.
> Strange...
> Anyway, it crashes when it tries to use a function that is not linked.
> I would have expected the linker to complain, and the compiler to at  
> least warn, but maybe I'm too used to "too helpful" C++ compilers...
> 
> The prototype for init_feature_variables, in  
> extract_block_features.c, changed from using one parameter, to using  
> four, in revision 1.6, committed more than a month ago... Is anyone  
> actually using the latest version?
> 
> With a segfaulting gift-write-feature-descs, the  
> InvertedFileFeatureDescription.db file is always empty...
> Saves some space on the harddisk of course, but other than that it  
> couldn't be all that good, right?
> 
> Does the GIFT work at all now?
> 
> Arghh...
> Now it would have been nice to have SVN changesets, so that reverting  
> a commit would revert all files that was committed...
> 
> /Jonas
> 
> 



> _______________________________________________
> help-GIFT mailing list
> [email protected]
> http://lists.gnu.org/mailman/listinfo/help-gift