[GRASS-dev] [GRASS GIS] #2438: Vect_get_line_type cannot handle line out of range
GRASS GIS
trac at osgeo.org
Fri Oct 24 00:08:33 PDT 2014
#2438: Vect_get_line_type cannot handle line out of range
--------------------------+-------------------------------------------------
Reporter: artegion | Owner: grass-dev@…
Type: defect | Status: reopened
Priority: normal | Milestone: 7.0.0
Component: LibVector | Version: unspecified
Resolution: | Keywords:
Platform: Unspecified | Cpu: Unspecified
--------------------------+-------------------------------------------------
Changes (by mmetz):
* milestone: 6.4.5 => 7.0.0
Comment:
Replying to [comment:2 artegion]:
> Replying to [comment:1 mmetz]:
> > Replying to [ticket:2438 artegion]:
> > > calling Vect_get_line_type with line out of range (i.e. line=0)
> > > Vect_line_alive returns -1 but Vect_get_line_type ignores it (not -1
is False just like not 1) -> access violation reading ....
> >
> > This is a programmer's error. 'Vect_get_line_type()' assumes that the
programmer first tested if the provided line id is indeed valid. You did
call 'Vect_line_alive()' first, you can only proceed if the line is alive.
Fix your code.
>
> For sure calling 'Vect_get_line_type()' with a wrong line id is a
programmer error but 'Vect_get_line_type()' already call
'Vect_line_alive()'.
>
>
> Testing if (!Vect_line_alive(Map, line)) is wrong because
Vect_line_alive returns[[BR]]
>
> 1 feature alive[[BR]]
>
> 0 feature is dead[[BR]]
>
> -1 on error[[BR]]
>
>
> a small change ( if (Vect_line_alive(Map, line!=1)) may improve code
robustness.
The problem comes from r55582 which introduced return -1 on error for the
Vect_*_alive functions. Unfortunately, the existing calls to
Vect_*_alive() were not adjusted. I have restored the behaviour of the
Vect_*_alive() functions in r62370,1, no more need to change
Vect_get_line_type(). Note that this applies only to G7, not to G6.
>
>
>
--
Ticket URL: <http://trac.osgeo.org/grass/ticket/2438#comment:4>
GRASS GIS <http://grass.osgeo.org>
More information about the grass-dev
mailing list