Discuss Tag:natural=coastline deprecation here
moved from Tag:natural=coastline
I've just re-jiggled this page to reflect a more current status. The tag is actually accepted onto the Map Features page.
But I've also tried to frame a discussion which needs to be had. Does someone actually want to propose that this tag be deprecated (and possibly database purged) ? Based on mailing list posts it seems several people think that coastline data should not be in the OpenStreetMap database.
Personally I'd favour keeping it, and continuing with imports, but maybe I'm not aware of all the issues. How about listing advantages and disadvantages on there?
I suspect other people are oblivious to this discussion (if indeed it is still under discussion) and are still following the instructions elsewhere to import new sections of coastline
-- Harry Wood 12:52, 14 August 2007 (BST)
- I wonder if there might not be a "middle way". I declare an interest here, I imported about half of the UK coastline, then tidied it up for all obvious errors due to problems with the shapefiles. I then used the Landsat imagery to better approximate the real coastline, and have been doing some further tidying up using Yahoo imagery now it is available. The raw PGS import is just not suitable for OSM final use and needs editing. However, as long as some similar tools weer available which allowed the raw PGS data to be edited and corrected, then the major reason for having the data in the main database is removed. Having said that it is nice to have all the data in one place so that other uses of all oSM data (and not just the Mapnik / tiles at home renderers) can easily use it. Dmgroom 10:53, 31 August 2007 (BST)
Yeah the issues around current renderer behaviour are less important than the long-term aims I suppose. I'm thinking we surely want to allow somebody to download a bbox with all the data they need to draw a map. In an area such as a coastal city, it seems a shame to provide detailed street-by-street postbox-by-pub data in a single easy download... but minus the coastline.
It's true we could devise a "middle way". In fact the status quo at the moment is a sort of middle way. We build natural:coastline ways in areas poeple are most interested in, which will include detailed coastline in many coastal cities. We avoid expanding the coast data too quickly (hold off from running bulk imports), because we don't want it to dwarf the size of the streets data, and cripple the server, but as people map more and more streets, we will also be adding more and more coastline, and eventually (several years down the line) we'll need the database to scale to the whole world somehow, for both street and coastline data.
Maybe the only problem with the current "middle way", is that it's not clear what the recommendations are regarding adding coastlines.
-- Harry Wood
: I'd add another problem with the current situation - Mapnik doesn't use the natural=coastline data at all, so many people looking at the slippy map may never see the corrected coastline where it exists. Dmgroom 14:22, 31 August 2007 (BST)
The whole deprecation idea was raised long before coastlines were even visible on the tiles@home layer and at the time it was unclear if rendering the sea according to imported coastline was even fesaible. It has meanwhile been proven that coastline can be used to make nice maps. Coastline is partly imported automatically from PGS etc., partly in a semi-automatic fashion from landsat images (lakewalker plugin), partly by manual trace from Yahoo images. There is no good reason to discard that - it works fine. I have only just seen that this discussion is alive here, I thought it was dead long ago. Removing coastlines from our database, for me, is not an option. --Frederik Ramm 15:00, 13 September 2007 (BST)
- Yeah OK. Sorry for the confusion. I resurrected this discussion (made the inconsistency more prominent) but it seems any talk of deprecation and purging was actually dead and gone a long time ago as you say. The concensus is not deprecated. Everyone I've spoken to seems to concurr. So just now I've stated that more clearly on the information here. Please check what I've written and feel free to tweak the wording.
- -- Harry Wood 10:33, 14 September 2007 (BST)
note all that discussion is quite old actually