[Magick-bugs] mogrify(1) bug in calling FT_Outline_Decompose()...

[Sean Chittenden]

>> That said, this regression is not caught in the regression tests, but
>> is reproducible on Solaris with the mogrify(1) command included
>> earlier.  Running the same mogrify(1) command on Linux of FreeBSD  
>> does
>> not result in a crash.
>
> This looks like an illegal NULL pointer reference.


Correct.  Something's setting outline->contours to an integer value  
vs. a pointer.  I spent plenty of time in gdb(1) but wasn't able to  
dig up where this was happening.  n_contours gets set to 1, and  
contours ends up having the value 12 (0xc) set to it.  :-/  When using  
a TTF font, the problem went away.  A PCF font, however, caused things  
to crash (unable to correctly obtain the glyph bounding box?  Just  
guessing at this point).  -sc

--
Sean Chittenden
sean at chittenden.org