[GRASS-dev] Re: georectifier fixes for Debian

[Otto Dassau]

On Fri, 3 Nov 2006 20:00:38 +1300
Hamish <hamish_nospam at yahoo.com> wrote:

> Michael Barton wrote:
> > I got a student here with Debian Ubuntu (Mepis) to try the --ui switch
> > for the i.group command in georectify.tcl. It seems to solve the
> > problem of i.group flashing on and off.
> 
> I have just updated to the latest CVS and the georectifier is now
> working properly (AFAIU), running i.group from the other side of the
> looking glass & all.
> 
> great!
> 
> > Thanks very much Hamish for suggesting it. It's in the cvs now. I'm
> > copying Markus so that he can backport it to 6.2.1
> 
> in 6.2.0 the "2. create a group" button now is a no-op for me (note I'm
> currently at 100% cpu + 80% RAM for a modelling task, so race-for-
> resource errors might be different). Not a big deal, but we'll have to
> answer the question of how to set the group up in another grass session
> eventually. If this is the only last-minute 6.2.0 breakage I'll be very
> happy.
> 
> 
> Otto:
> > > the i.group window opens, but when I close the "error" message below
> > > appears, but I can still go on rectiifying.
> > > 
> > > child process exited abnormally
> > > child process exited abnormally
> > >      while executing
> > >  "exec -- $cmd --ui"
> > >      (procedure "GRMap::group" line 17)
> > >      invoked from within
> > >  "GRMap::group"
> > >      ("uplevel" body line 1)
> > >      invoked from within
> > >  "uplevel \#0 $cmd"
> > >      (procedure "Button::_release" line 18)
> > >      invoked from within
> > >  "Button::_release .grstart.mf.frame.group.a"
> > >      (command bound to event)
> > >  and set GCPs
> 
> No harm in it, AFAIK, just normal termination(?). The "catch" statement
> hides this error message. Unfortunately the "catch" was also catching a
> genuine error (expecting command line options; thus it needs a --ui GUI
> [why is that?]).

ok, thanks a lot for your help and fixes. I tested the current cvs version Grass
6.3 head and the georectifier is working very well now - great!!

thanks a lot

 Otto

> From what I've heard here the Tcl return test is bogus so there is no
> way to test a good exit from a bad one?
> 
> perhaps gis.m could use a global DEBUG 0/1 setting like the start of the
> nviz2.2_script has. With it changed to 1 it could disable the catches,
> etc. or maybe better it could test on startup if the gisenv DEBUG was >
> 0 to set a global DEBUG var.
> 
> 
> Hamish
> 


--