[GRASS5] Re: G_fatal_error

[Markus Neteler]

Hi again,

the G_fatal_error seems to be detected:

Roger wrote:
 I've now narrowed my G_fatal_error() problem down to G_gisinit(); before                         
 G_gisinit() is called, you can't set the handler function. After                                 
 G_gisinit(), I can reset it to use the R error() and warning() functions,                        
 jumping back out of interface code to the R prompt safely for fatal                              
 errors. For now I'm checking environment variables before calling                                
 G_gisinit, but the longer-term solution is to do an interpreted                                  
 system("g.whatever") and test the value returned by exit(). If the GRASS                         
 environment is intact and the user mapset is writable, that should be                            
 enough. The functions in G_gisinit() get to call so many functions which                         
 fail with G_fatal_error() that trying to guard against them all isn't                            
 possible. Once past G_gisinit(), we're OK.

Maybe this info is useful for you...

Cheers

 Markus
-------------- next part --------------
An embedded message was scrubbed...
From: Roger Bivand <rsb at reclus.nhh.no>
Subject: Re: GeoComputation 2000 (fwd)
Date: Fri, 16 Jun 2000 12:20:43 +0000 (   )
Size: 2128
Url: http://lists.osgeo.org/pipermail/grass-dev/attachments/20000616/dc0b12b1/attachment.mht