Re: [GRASS-user] v.edit tool=merge

[Vincent Bain]

Le lundi 26 septembre 2016 à 09:38 +0200, Markus Metz a écrit :

> Maybe coordinates changed slightly during export, this could explain
> why we get different results. Line heights for the same contour are
> not exactly equal, e.g. 1239.39 for one line and 1240 for the next
> line.

Yes indeed, there was a little mess in my source data, inherited from a
transfer from local to global coordinate system.

I know I should give up using G64 but (sorry for not formalizing more my
position) I still make a quite heavy use of it when it comes to
intensive digitizing tasks. I consider the digitizer is more stable and
faster for instance when displaying raster backgrounds (again I am
guilty to complain without pushing up issues, it's just a side remark).

Thanks again,
Vincent


___
grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user

Re: [GRASS-user] v.edit tool=merge

[Markus Metz]

On Sun, Sep 25, 2016 at 8:36 AM, Vincent Bain  wrote:
> Le samedi 24 septembre 2016 à 22:31 +0200, Markus Metz a écrit :
>
>> These zero-length lines should not cause an infinite loop, fixed in
>> all G7 branches with r69569-71. With these changes, you also need only
>> one run of v.edit in order to merge all selected lines.
>
> Yep' perfect!
>>
>> With G64, v.edit tool=merge hangs for me on that zero-length line.
>> Weird, that G64 can successfully merge these lines for you.
>
> well, my last built of 6.4.x release dates back to mid august.

the part of G64 relevant for v.edit tool=merge last changed more than
8 years ago.

Maybe coordinates changed slightly during export, this could explain
why we get different results. Line heights for the same contour are
not exactly equal, e.g. 1239.39 for one line and 1240 for the next
line.

Markus M

>
> Vincent.
>
>
___
grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user

Re: [GRASS-user] v.edit tool=merge

[Vincent Bain]

Le samedi 24 septembre 2016 à 22:31 +0200, Markus Metz a écrit :

> These zero-length lines should not cause an infinite loop, fixed in
> all G7 branches with r69569-71. With these changes, you also need only
> one run of v.edit in order to merge all selected lines.

Yep' perfect!
> 
> With G64, v.edit tool=merge hangs for me on that zero-length line.
> Weird, that G64 can successfully merge these lines for you.

well, my last built of 6.4.x release dates back to mid august.

Vincent.


___
grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user

Re: [GRASS-user] v.edit tool=merge

[Markus Metz]

On Sat, Sep 24, 2016 at 6:17 PM, Vincent Bain  wrote:
> Le vendredi 23 septembre 2016 à 23:11 +0200, Markus Metz a écrit :
>> Maybe it is a zero-length line issue. You can remove these zero-length
>> lines with v.clean tool=rmline and check if editing the cleaned vector
>>  still causes these GUI crashes.
>
> You're right : cleaning the map prior to editing prevents the crash.

These zero-length lines should not cause an infinite loop, fixed in
all G7 branches with r69569-71. With these changes, you also need only
one run of v.edit in order to merge all selected lines.

With G64, v.edit tool=merge hangs for me on that zero-length line.
Weird, that G64 can successfully merge these lines for you.

Markus M
___
grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user

Re: [GRASS-user] v.edit tool=merge

[Vincent Bain]

Le vendredi 23 septembre 2016 à 23:11 +0200, Markus Metz a écrit :
> Maybe it is a zero-length line issue. You can remove these zero-length
> lines with v.clean tool=rmline and check if editing the cleaned vector
>  still causes these GUI crashes.

You're right : cleaning the map prior to editing prevents the crash.

Merci!
Vincent.

___
grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user

Re: [GRASS-user] v.edit tool=merge

[Markus Metz]

On Fri, Sep 23, 2016 at 10:27 PM, Vincent Bain  wrote:
> Le vendredi 23 septembre 2016 à 22:17 +0200, Markus Metz a écrit :
>> At least you got a working solution.
>
> Yes, thanks !
>
> working with the same dataset, I ran into several GUI crashes (no
> time/no way to get an error message cause the window shuts up
> immediately, perhaps there's a log file but I don't know where) when
> trying to edit the vector layer. In the source vector file (extracted
> from a blender file, anyway) I purged two double vertices, reimported
> it, then digitizing went fine in the end. It sounds like a zero-length
> line issue again, no ?

Maybe it is a zero-length line issue. You can remove these zero-length
lines with v.clean tool=rmline and check if editing the cleaned vector
 still causes these GUI crashes.

Markus M
___
grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user

Re: [GRASS-user] v.edit tool=merge

[Vincent Bain]

Le vendredi 23 septembre 2016 à 22:17 +0200, Markus Metz a écrit :
> At least you got a working solution.

Yes, thanks !

working with the same dataset, I ran into several GUI crashes (no
time/no way to get an error message cause the window shuts up
immediately, perhaps there's a log file but I don't know where) when
trying to edit the vector layer. In the source vector file (extracted
from a blender file, anyway) I purged two double vertices, reimported
it, then digitizing went fine in the end. It sounds like a zero-length
line issue again, no ?

Best,
Vincent.

___
grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user

Re: [GRASS-user] v.edit tool=merge

[Markus Metz]

On Fri, Sep 23, 2016 at 5:44 PM, Vincent Bain  wrote:
> Thanks Markus,
> indeed it works much much faster !
>
> However on my data I notice the result is slightly different from the
> one I get with  6.4, i.e. some segments don't snap to each other : where
> I should have a result with 199 lines I have 217 objects. If I try to
> snap them manually via wx digitizer, it fails too.
> Later I noticed that when applying twice the command, the result is
> finally correct (199 lines).

Weird. I will try to look into it, but this is will take some time.
For the current fix, I needed to go through 600 MB of debug messages.
The difference between G7 and G6 is a mystery to me, and I don't
understand (yet) why it works with G6 at all. I guess it is a
difference somewhere in the main vector library, problem is that the
G6 and G7 vector libraries are very different.

At least you got a working solution.

Markus M

>
> Vincent.
>
> Le vendredi 23 septembre 2016 à 15:00 +0200, Markus Metz a écrit :
>> On Tue, Sep 20, 2016 at 3:06 PM, Vincent Bain  wrote:
>> > Hi,
>> > on a fresh (r69498) trunk version of grass, I have a problem running a
>> > v.edit command on a map.
>> > Given a vector map composed of 200 contour lines splitted in roughly
>> > 17000 line segments (adjacent nodes perfectly overlapping), I need to
>> > "unsplit" these contours. If I try to run the following command :
>> >
>> > v.edit map=contours type=line tool=merge snap=vertex -r cats=0
>> >
>> > it looks line it never ends, i.e. the process does not return any error,
>> > RAM and CPU usages appear stable (not huge). After tens of minutes it
>> > still hangs up. I finally have to interrupt, then chucking out the
>> > topology-damaged map.
>>
>> v.edit tool=merge hangs for zero-length lines. Such contours can occur
>> on peaks and in pits. Fixed in all G7 branches with r69556-8.
>>
>> Markus M
>>
>> >
>> > Running the same command on grass6.4 (after v.build the source map), it
>> > runs successfully within some minutes.
>> >
>> > Before I open a ticket, would anyone wish to give a try to my data ?
>> >
>> > Vincent.
>> >
>> > ___
>> > grass-user mailing list
>> > grass-user@lists.osgeo.org
>> > http://lists.osgeo.org/mailman/listinfo/grass-user
>
>
___
grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user

Re: [GRASS-user] v.edit tool=merge

[Vincent Bain]

Thanks Markus,
indeed it works much much faster !

However on my data I notice the result is slightly different from the
one I get with  6.4, i.e. some segments don't snap to each other : where
I should have a result with 199 lines I have 217 objects. If I try to
snap them manually via wx digitizer, it fails too.
Later I noticed that when applying twice the command, the result is
finally correct (199 lines).

Vincent.

Le vendredi 23 septembre 2016 à 15:00 +0200, Markus Metz a écrit :
> On Tue, Sep 20, 2016 at 3:06 PM, Vincent Bain  wrote:
> > Hi,
> > on a fresh (r69498) trunk version of grass, I have a problem running a
> > v.edit command on a map.
> > Given a vector map composed of 200 contour lines splitted in roughly
> > 17000 line segments (adjacent nodes perfectly overlapping), I need to
> > "unsplit" these contours. If I try to run the following command :
> >
> > v.edit map=contours type=line tool=merge snap=vertex -r cats=0
> >
> > it looks line it never ends, i.e. the process does not return any error,
> > RAM and CPU usages appear stable (not huge). After tens of minutes it
> > still hangs up. I finally have to interrupt, then chucking out the
> > topology-damaged map.
> 
> v.edit tool=merge hangs for zero-length lines. Such contours can occur
> on peaks and in pits. Fixed in all G7 branches with r69556-8.
> 
> Markus M
> 
> >
> > Running the same command on grass6.4 (after v.build the source map), it
> > runs successfully within some minutes.
> >
> > Before I open a ticket, would anyone wish to give a try to my data ?
> >
> > Vincent.
> >
> > ___
> > grass-user mailing list
> > grass-user@lists.osgeo.org
> > http://lists.osgeo.org/mailman/listinfo/grass-user


___
grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user

Re: [GRASS-user] v.edit tool=merge

[Markus Metz]

On Tue, Sep 20, 2016 at 3:06 PM, Vincent Bain  wrote:
> Hi,
> on a fresh (r69498) trunk version of grass, I have a problem running a
> v.edit command on a map.
> Given a vector map composed of 200 contour lines splitted in roughly
> 17000 line segments (adjacent nodes perfectly overlapping), I need to
> "unsplit" these contours. If I try to run the following command :
>
> v.edit map=contours type=line tool=merge snap=vertex -r cats=0
>
> it looks line it never ends, i.e. the process does not return any error,
> RAM and CPU usages appear stable (not huge). After tens of minutes it
> still hangs up. I finally have to interrupt, then chucking out the
> topology-damaged map.

v.edit tool=merge hangs for zero-length lines. Such contours can occur
on peaks and in pits. Fixed in all G7 branches with r69556-8.

Markus M

>
> Running the same command on grass6.4 (after v.build the source map), it
> runs successfully within some minutes.
>
> Before I open a ticket, would anyone wish to give a try to my data ?
>
> Vincent.
>
> ___
> grass-user mailing list
> grass-user@lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/grass-user
___
grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user