[Qgis-developer] New GRASS plugin: a test drive

[Radim Blazek]

On Sat, Oct 17, 2015 at 8:47 AM, m roy <royroge at outlook.com> wrote:
>
>
> Il 15/10/2015 13.25, Radim Blazek ha scritto:
>> On Thu, Oct 15, 2015 at 10:49 AM, m roy <royroge at outlook.com> wrote:
>>
>>> After some quick vector digitizing tests: awesome tool
>>> for QGIS !!
>>>
>>> the only (minor) issue i found is that area boundary color does not
>>> match the simbology settings (e.g. green boundary also if it's not
>>> shared by two areas)
>> Green means that both sides are topologically correct. If there is no
>> area outside a ring, it is correct. Try to attach a dangling boundary
>> to the ring and it should become orange.
>> Maybe the name in symbology is not precise.
>>
>> Radim
>
> I think it's better to stick with GRASS symbology convention,
> otherwise it's quite confusing if you use both systems;
>
> e.g. consider the *isle* below
>
> ++++++o++++++
> +          *          +
> +          *          +
> ++++++o++++++
>
> + area on one side (of the boundary) e.g. orange
>
> * area on both sides, e.g. green

In the meantime, I have divided boundary with area on one side into
two symbols, so that it is possible to show visually which side is
broken. The style is left the same for now, because I am not sure
which is the best (parallel lines with different color, red markers
with offset to broken area...) but user can set his own style. I have
also renamed the symbols:
  * Boundary (topological error on both sides
  * Boundary (topological error on the left side)
  * Boundary (topological error on the right side)
  * Boundary (correct)

I believe that if a boundary is topologically incorrect, even on one
side and even if the chain is broken in another place,  it should be
visualized. Your argument about consistency with GRASS is important,
but in this case, I'll probably resist.

BTW, try this in GRASS:

+++++++++++++++++++
+                                    +
+    ++++++o++++++     +
+    +          *            +    +
+    +          *            +    +
+    ++++++o++++++     +
+                                    +
+++++++++++++++++++

Does it still make sense for the boundaries of inner areas?

> there seems to be some problem with the symbology while
> in edit mode, some how all boundaries disappear at some
> point, but if you exit edit mode and save everything is ok

New boundaries digitized in this session or old boundaries which were
present in the map before editing started or both. Are you able to
describe "at some point" with reproducible steps? If it happens again,
try to look into attribute table and check topo_symbol, select a
record and check if it is highlighted in the map and open attribute
form and check topo symbol there.

Thanks
Radim