I agree with James, until we have mapped all trees as nodes (!-) removing 
current forest is not a good idea.
OSM works that way… you have some free time, you add/correct the data you can. 
Without doing this, the map would simply be empty (and we couldn’t see the 
white rectangles we are talking about!)

Daniel


From: James [mailto:james2...@gmail.com]
Sent: Tuesday, 30 August, 2016 06:13
To: Gordon Dewis
Cc: Antoine Beaupré; Talk-CA OpenStreetMap
Subject: Re: [Talk-ca] broken forests in eastern Canada


Yeah I have to run JOSM in 64 bit mode via the jnlp to manage such massive data 
amounts as the 32bit is slower and is limited about 1.4-1.6GB on windows and 
2GB on linux because of how the heap works.

P.s. if someone feels up to it we could always map trees as node=tree

On Aug 29, 2016 11:49 PM, "Gordon Dewis" 
<gor...@pinetree.org<mailto:gor...@pinetree.org>> wrote:

On Aug 29, 2016, at 11:12 PM, Antoine Beaupré 
<anar...@orangeseeds.org<mailto:anar...@orangeseeds.org>> wrote:

On 2016-08-25 10:13:25, Gordon Dewis wrote:

Alan is right. I've brought in a few tiles worth of forests from Canvec in
the area you're talking about, but they were non-trivial to deal with
compared to most other features. I kept running into limits in the tools I
was using at the time and I haven't returned to them since.

Yeah, that's what I figured.... I hope my comment didn't come across as
criticizing the work that was done importing that data into OSM - I know
how challenging and frustrating that work can be.

But I must admit it seems a little rough to have those patches up
there. I don't mind the "seams" between the CANVEC imported blocks,
which don't seem to show up on the main map anymore anyways. But
the *missing* blocks are really problematic and confusing. And they show
up not only all the way up north and in weird places, but in critical
areas. for example, here's a blank spot right north of Canada's capital:

http://osm.org/go/cIhYCSU-?m=

It seems a whole area was just not imported up there... oops! This shows
up here and there in seemingly random places.

Whoever was working on it was probably struggling with the tiles and subtitles 
in Canvec and threw in the towel. I was working on the forests around Golden 
Lake, for example, and ran into problems and limitations with the tools I was 
using at the time. I would love to import more, but it’s a daunting task.

Another problem I noticed is when trying to merge “new” forests with existing 
forests was the existing forests would disappear because the topology changed, 
similar to problems you can see with lakes and islands. That alone was enough 
to make me back off and undo the inadvertent damage.

I wonder if it wouldn't be better to remove parts of the CANVEC import
until we can figure out how to better import them in the future, if, of
course, we have a documented way of restoring the state of affairs we
have now... As was mentionned elsewhere, it seems to me that the data
that is there now somewhat makes it more difficult to go forward and
hides more important data (like park boundaries).

Unless the parts of Canvec are going to be replaced with more comprehensive 
coverage, I think that removing the existing forests would not be a Good Thing.


I believe it would be more important to map out park boundaries than
actual forest limits which, quite unfortunately, change in pretty
dramatic ways in Québec, due to massive logging that has been happening
for decades.

Park boundaries are mostly in already, aren’t they? They are fairly easy 
features to import compared to forests.

_______________________________________________
Talk-ca mailing list
Talk-ca@openstreetmap.org<mailto:Talk-ca@openstreetmap.org>
https://lists.openstreetmap.org/listinfo/talk-ca
_______________________________________________
Talk-ca mailing list
Talk-ca@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-ca

Reply via email to