Metric distance option for v.distance in latlon regions

[cmbarton]

It would be nice if v.distance could calculate distance (e.g. minimum distance between to and from features) in a metric value like meters even when it works in a latlon region. Currently, it only calculates distance in the default units of the region.

Michael

[hamish]

if so it shouldn't be too hard to get, G_geodesic_distance() is always returning meters, so we already know the number internally.

since it isn't a bugfix, I suggest not to add new flags in the 6.4.3 release branch.

Hamish

[cmbarton]

You're right about it not being a bug fix. I'll bump it up to 6.4.4 (or is 6.5 and above better?) and above.

Michael

[hamish]

Replying to cmbarton:

You're right about it not being a bug fix. I'll bump it up to 6.4.4 (or is 6.5 and above better?) and above.

I guess the normal procedure would be to add it to trunk first, and then see how invasive the change was before considering backporting it to 6.x. If destined for 6.4.4 for sure and the 6.4.3 branch is frozen for release as it is now, I'd backport from trunk to devbranch_6 and then (and I'm not very good at remembering this part) make sure to add it to the list of stuff to backport in the trac dev wiki pages so it doesn't get forgotten once the release branch 64 is reopened post release.

Hamish

[cmbarton]

All good thoughts. The question at hand is what to set the choice box at. I've moved it to 6.5. Makes sense to try it in 7 first, though.

[mmetz]

Replying to hamish:

if so it shouldn't be too hard to get, G_geodesic_distance() is always returning meters, so we already know the number internally.

This does not work for v.distance. You would need to enhance G_distance_point_to_line_segment() with the options of Vect_line_distance() or write a new function Vect_line_geodesic_distance().

Bumping Milestone.

Markus M

[cmbarton]

So it hasn't yet been done because it is more work than it initially seemed. Well maybe it can happen for GRASS 7. Not an urgent thing, but it would be nice...

Michael

[hamish]

ah, calculating the great circle distance between every node of a polyline could be rather computationally expensive, but it is possible if you don't mind the wait.

to clarify: does the current version report decimal degrees (aka garbage) for v.distance in a lat/lon location? If so it at least deserves a warning, and in general a fatal error is better than plausible but actually meaningless results.

Hamish

[cmbarton]

Yes. It reports in decimal degrees. Not too useful.

Michael

[martinl]

Milestone renamed

[martinl]

Still the issue?

[neteler]

Ticket retargeted after milestone closed

[cmbarton]

Fixed in both 7.2.2 and 7.4