Re: [Talk-us] Multipolygonizing

[Andy Townsend]

On 20/11/2017 23:30, Ian Dees wrote:


Please remember to stay on topic and friendly. This thread seems to be 
drifting off into a discussion about the merits of OSM editors.


Well, my comment about editors wasn't supposed to be offtopic, since the 
question of data being "... far easier to understand and maintain, 
especially for novice mappers" was one of the points raised at the very 
top of the thread.


It's perhaps worth mentioning that in each of iD, P2* and JOSM (without 
plugins) it's possible to swap without too much difficulty between the 
two relations and the constituent ways at 
http://www.openstreetmap.org/#map=19/36.62063/-121.90621  P2's internal 
visualiser fails with the park visualisation though, and I can't see a 
way to select the marine nature reserve without deliberately selecting 
the "relations this way is a member of" at the left, so I'm still not 
convinced that this area is as newbie-friendly as it was before.


Best Regards,
Andy


* if you are surprised by this perhaps you haven't looked at one or 
another editor for a while - it might be worth revisiting.


___
Talk-us mailing list
Talk-us@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-us

Re: [Talk-us] Multipolygonizing

[OSM Volunteer stevea]

On Nov 21, 2017, at 1:27 PM, Gleb Smirnoff  wrote:
> Okay, I will withhold myself from touching polygons in the Santa Cruz County
> for next couple of years, and let's see how your future experience with
> SCCGIS goes on. We can get back to this question later in scope of Santa Cruz.

This is a very happy result, thank you for the good (if rather public in 
talk-us) dialog.  I think it was beneficial to the greater OSM community that 
our dialog was so public, as Kevin and I have been discussing "shared ways in 
multipolygons vs. regular polygons" off-list for some time, and I've always 
known this trend towards "shared ways" would deeply affect a large import I 
keep updated in my county.  I believe this topic has made other OSM 
importers/maintainers of mostly- or exclusively-polygon data wonder what the 
best course of action is as OSM evolves to "shared ways" becoming more and more 
common.  I hope it has helped a better consensus to emerge – it feels like it 
is doing so locally.

What is emerging (at least here, between me and Gleb) is that there will come a 
point in either initial/original imports that are largely or exclusively 
polygon-only when it simple becomes time to "bite the bullet" and do the 
initial work to convert these to multipolygon as the trend towards "shared 
ways" grows.  Yes, that is lots of work up front, but I believe in advance that 
it will be worth it in the editing time/efforts saved in the future as Gleb and 
Kevin have pointed out its many editing benefits.  (I agree it is easier to 
maintain such "edges," boundaries especially, including landuse, which are 
"shared ways" as multipolygons allow us to do.  EXCEPT in large, existing 
imports!).

> Meanwhile, do I understand that my initial understanding of strong consensus
> against multipolygons in the USA overall was wrong reading? First few emails
> in the thread made me think so.

Gleb, it was a sort of misunderstanding, and it doesn't seem important to lay 
blame on anybody in particular.  What is important is that we seem to agree 
that while polygons certainly have their place and aren't going away, 
multipolygons are here to stay as well, and there is a distinct trend towards 
using them in a "shared way" context where it makes sense to do so.  (The good 
examples that Kevin listed, likely more).  Yes, as Frederik said, it can be a 
matter of taste which is better, as both are correct (one is harder to edit in 
one context, the other is harder to edit in another context), and so we should 
not be spending time "converting" from polygons to multipolygons.  However, 
where it makes sense to use multipolygons in NEW data, let us enter them as 
such.

> I'd like to continue working on coastline, and map all remaining SMRs and
> later maintain them. I also want keep using multipolygons in any regular
> edits. Are there any objections?

If by "regular edits" you mean adding NEW data, no, I have no objection.  If 
you want to "convert" existing polygons to multipolygons, yes, I (and others) 
object.

Thank you once again for good, productive dialog!

SteveA
California
___
Talk-us mailing list
Talk-us@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-us

Re: [Talk-us] Multipolygonizing

[Joel Holdsworth]

On 21/11/17 14:29, Gleb Smirnoff wrote:

   Steve,

On Mon, Nov 20, 2017 at 04:34:18PM -0800, OSM Volunteer stevea wrote:
O> If the reltoolbox plug-in as as powerful as I am beginning to understand it 
may be (I appreciate the introduction, Gleb), and given my agreement that certain 
use cases (especially landuse) benefit greatly from multipolygonized boundaries 
(they do), I actually CAN imagine that the SCCGIS V4 landuse import data (in 2019 
or 2020) could become multipolygon.  This likely would involve a pre-upload 
translation of polygon data into mulitipolygon using the tool, then conflation 
(which has to be done anyway).  Except, we upload multipolygons as we delete 
existing polygons during the conflation-and-upload phase.
O>
O> I wanted to offer that bright spot of hope to anybody's lingering beliefs that I am 
"mule-entrenched" in my beliefs that existing polygons are always superior.  They are not.  
They make updates harder, but I think I can get over that, as I can be convinced that "once done, 
the time investment is worth it" for the future benefits that multipolygons bring.

Okay, I will withhold myself from touching polygons in the Santa Cruz County
for next couple of years, and let's see how your future experience with
SCCGIS goes on. We can get back to this question later in scope of Santa Cruz.

Meanwhile, do I understand that my initial understanding of strong consensus
against multipolygons in the USA overall was wrong reading? First few emails
in the thread made me think so.

I'd like to continue working on coastline, and map all remaining SMRs and
later maintain them. I also want keep using multipolygons in any regular
edits. Are there any objections?



I use multipolygons extensively for the land cover around Rocky Mountain 
National Park.


___
Talk-us mailing list
Talk-us@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-us

Re: [Talk-us] Multipolygonizing

[Gleb Smirnoff]

  Steve,

On Mon, Nov 20, 2017 at 04:34:18PM -0800, OSM Volunteer stevea wrote:
O> If the reltoolbox plug-in as as powerful as I am beginning to understand it 
may be (I appreciate the introduction, Gleb), and given my agreement that 
certain use cases (especially landuse) benefit greatly from multipolygonized 
boundaries (they do), I actually CAN imagine that the SCCGIS V4 landuse import 
data (in 2019 or 2020) could become multipolygon.  This likely would involve a 
pre-upload translation of polygon data into mulitipolygon using the tool, then 
conflation (which has to be done anyway).  Except, we upload multipolygons as 
we delete existing polygons during the conflation-and-upload phase.
O> 
O> I wanted to offer that bright spot of hope to anybody's lingering beliefs 
that I am "mule-entrenched" in my beliefs that existing polygons are always 
superior.  They are not.  They make updates harder, but I think I can get over 
that, as I can be convinced that "once done, the time investment is worth it" 
for the future benefits that multipolygons bring.

Okay, I will withhold myself from touching polygons in the Santa Cruz County
for next couple of years, and let's see how your future experience with
SCCGIS goes on. We can get back to this question later in scope of Santa Cruz.

Meanwhile, do I understand that my initial understanding of strong consensus
against multipolygons in the USA overall was wrong reading? First few emails
in the thread made me think so.

I'd like to continue working on coastline, and map all remaining SMRs and
later maintain them. I also want keep using multipolygons in any regular
edits. Are there any objections?

-- 
Gleb Smirnoff

___
Talk-us mailing list
Talk-us@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-us