[OSM-dev] Very long ways have been split (was: Status of Database Server after 0.4 Upgrade: Fragile)

[Frederik Ramm]

Hi,

> Please remember the Mapnik layer too -- that will only render areas in
> solid colour if they are a single closed way. Until now mapnik has never
> had an issue with such long ways since it has been using the planet dump
> which has no arbitrary limits on way length or bounding boxes. 

Hm. But islands do work even if they do not have the coastline as one 
single way, don't they?

So then the controversy would boil down to those 19 of 249 ways that 
were tagged natural=water, i.e. lakes (one should think - Donau at least 
is the German name for the Danube River...).

A closer look at those gives:

3791403 Balaton (889 segs)
3868897 (959 segs)
3870479 (2216 segs)
3870598 (714 segs)
3870625 Donau (861 segs)
3870656 (1531 segs)
3871262 Donau (813 segs)
4018578 (737 segs)
4049723 (1214 segs)
4101887 (1133 segs)
4198853 Lake Burley Griffin (1238 segs)
4272451 (506 segs)
4279603 (3842 segs)
4289157 Lago Maggiore (1902 segs)
4306842 (622 segs)
4374275 Inya Lake Inya Lake Victoria Lake (651 segs)
4397275 Ghost Dam (547 segs)
4547927 Bodensee (872 segs)
4548813 (594 segs)

Now would you (not you=Jon, but you=all of you) suggest? Should I 
re-instate all of them, potentially upsetting the API (and the 
unsuspecting JOSM user) with 3842 segments in one way? I believe there 
will be a lot of lakes in Canada and Russia that would require even more 
than 3800 segments to be drawn properly, and trying to say "one way=one 
area" seems to have no future.

My personal opinion would be to accept that splitting large areas is a 
requirement and renderers have to deal with this one way or another, and 
I was just about to start fixing tiles at home to not only accept split 
coastlines, but also other split area boundaries.

That said, if you'd rather have a rollback on this, I can provide it.

Bye
Frederik

-- 
Frederik Ramm  ##  eMail frederik at remote.org  ##  N49°00.09' E008°23.33'