[GRASS-user] v.edit bus error

[Glynn Clements]

William Kyngesburye wrote:

> >> That's probably the same bug I just found in another module, and is
> >> present in v.edit also.  Bug already filed, and it's in the process
> >> of being fixed in CVS.
> >
> > Which one?
> 
> http://wald.intevation.org/tracker/?func=detail&atid=204&aid=455&group_id=21

Not related.

> > Note that this is a bus error, not a segfault. On PPC, the most common
> > reason for a bus error is unaligned access (e.g. trying to read a
> > 32-bit word from a memory address which isn't a multiple of 4).
> >
> > x86 allows unaligned access, so this kind of bug normally only gets
> > discovered on Macs.
> 
> Intel OSX.  I don't know what the default alignment is for Intel  
> OSX.  The GCC manpage wasn't clear what the Intel defaults are for  
> OSX (or it's clear in a complex way).

Intel doesn't care about alignment; it will just read both words and
piece the value back together (a consequence of its 8-bit origins).

The link in the original message refers to PPC binaries.

I can't see anything obvious in v.edit itself, so tracking this down
depends upon someone providing a backtrace.

-- 
Glynn Clements <glynn at gclements.plus.com>