[GRASS-user] r.reclass and "Abort trap"

[Hamish]

DGann wrote:
> I am a fairly new GRASS user who recently shifted to a Mac computer (Lion
> OS).  The program seems to run well on the new platform for the most part,
> although periodically a line of working code, when copied and pasted
> directly from the GUI (layer manager) to the command terminal, will
> produce an error message.  I will continue to get error messages until
> I reboot GRASS, at which time the same original line of code will
> work no problem.  Sometimes the same line of working code simply never
> works again, such as in this present case:

sounds a lot like a memory/pointer error in the grass code.


> The task is really quite simple: a simple reclass of two categories
> into 1. 
> My rules .txt file looks like this (or some variation of this with or
> without different amounts of spacing):
> 1 2 = 1
> 
> Therefore categories 1 and 2 should be reclassed as category
> 1 (I have not attempted to give it a label yet).  Here is the
> command:
> 
> r.reclass --overwrite input=map1 at Forest_Attributes output=map1_reclass
> rules=temp.txt
> 
> The result is an empty map, and command terminal invariably says this:
> "Abort trap: 6"
> 
> I can't imagine the code is wrong, though I am open to suggestions....
>  I have searched the GRASS book and forums and there is precious little
> about any "Abort trap" output.  However I gather from the web
> that it's some sort of memory issue (buffer overflow)?

apparently the stack is being smashed.

Can you try 6.5svn? there is a segfault fix for r.reclass there (see
trac bug #800) which has not yet been backported to 6.4.


Hamish