[OSM-talk] Mapnik at zoom=19

the Old Topo Depot oldtopos at novacell.com
Sun Jan 27 17:41:18 GMT 2013


A quick extrapolation re space is to take rendered zl18 tile size, in
total, and multiply by 4.  If someone can provide the zl18 number ...

It seems reasonable to expect zl19 tiles to be of most interest in urban
areas with high POI density, as those areas tend to drop display data due
to labeling/feature densities.

While it is possible that a bot could cause all zl19 tiles to be rendered
and consume significant tile cache space, if this has not happened to date
on the existing tile server then it will likely not happen with zl19 either.

Be aware that code changes may be required to support higher zoom levels as
I have a vague, and quite likely outdated, recollection of arrays in mapnik
(?) and/or the rendering stack needing expansion.

On Sun, Jan 27, 2013 at 9:02 AM, Martin Koppenhoefer <dieterdreist at gmail.com
> wrote:

> 2013/1/27 Richard Fairhurst <richard at systemed.net>:
> >> Space requirements depend mostly on the actual usage (how many of them
> >> would be created). I guess it wouldn't make sense to prerender them
> >
> >
> > "I guess" is nice for mailing lists, but the sysadmins can't be asked to
> > make decisions based on a guess. Evidence is more helpful.
>
>
> in my tests zoom 19 took roughly the same time than the zoom 18 tiles
> but actual requirements (space and time) will depend mostly on how
> intensively people will use this feature on osm.org. This is hardly
> testable or predictable.
>
> cheers,
> Martin
>
> _______________________________________________
> talk mailing list
> talk at openstreetmap.org
> http://lists.openstreetmap.org/listinfo/talk
>



-- 
John Novak
585-OLD-TOPOS (585-653-8676)
http://www.linkedin.com/in/johnanovak/
OSM ID:oldtopos
OSM Heat Map: http://yosmhm.neis-one.org/?oldtopos
OSM Edit Stats:http://hdyc.neis-one.org/?oldtopos
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20130127/8f8f84a1/attachment.html>


More information about the talk mailing list