[OSM-dev] ...more OSM coming soon

[Joerg Ostertag (OSM Munich/Germany)]

On Monday 27 November 2006 10:32, Nicola Ranaldo wrote:
> On Wednesday 22 November 2006 12:05, Nick Black wrote:
> [...]
>
> > > > On 11/21/06, Nicola Ranaldo <ranaldo at unina.it> wrote:
> > > > > Please may someone update news about tile rendering?
> > > > > I'm waiting for about ten days and my area is still not ready...
>
> [...]
>
> > Rendering tiles takes time and processing power.  Why dont you render
> > some tiles of your area and upload them to your dev account ?.  From
> > there they can be very quickly and easily moved inot the tile
> > directory.
>
> I do not have a dev account, if i can obtain one i'll contribute to tile 
> rendering of Italy. 

You don't really need a dev account to contribute. I think I was too short. 
The dev account only was an easy way to hand over the rendered tiles to 
steve. SO sending an e-mail or putting it on a website is as fine as putting 
it on @dev.

> Howewer I think this has a sense only if the central 
> osm server will never process my area (or not before an acceptable time).

I think it will in (hopefully) acceptable time. Since there are already speed 
improvements in preparation, i hope we won't need this many rendering 
machines in the future. 

> Howewer in the next future, after the initial tile setup, we'll see tons of
> tiles rendering update requests! 

I think it'll depend on how many edits in an region have been done in one 
week(between 2 planet.osm) and if we can easily find out which tiles we 
really have to re-render. 

> If the number of indipendent renderers 
> will increase it will be hard to coordinate all them and put online their
> tiles.

Well it sure will be a little challenge to have this up and running. But as 
you can already see with tiles at home this would be something which can be made 
working.

> So i do not know if this is the rigth way... 

Well this distributed rendering was mainly for bringing up the tile cache to 
have rendered data in as many regions as possible. On the other hand it also 
gave a good test on performance, scalability and usability of the renderer. 
And, there is one more thing you can see if you look at different tiles which 
where rendered with different rules and software version. The Quality of the 
output improves with every input given from users and transformed to a new or 
improved rendering rule or programm. 

> The problem is the absence of news and roadmaps. 

It is probably very difficult or almost impossible to give a Roadmap in the 
current state of the project. As what I understood some here (me included) 
can't really tell in advance how much time of the day will be left for 
osm-specific stuff besides all the other things which have to be done. So you 
can't really count on a specific amount of time being spent for doing things 
which will bring us forward in the sense of a OSM-Roadmap.

But to try a vague prognosis of a Roadmap:
I think one of our next steps might be: To get the whole world rendered with 
mapnik out of a planet.osm in a few days. This could happen in one of the 
next weeks. After that we'll have to improve the rendering rules for the 
renderer.
But as you can already see mapnik is not the only renderer suitable for OSM. 
So there might arise also some other alternatives. Lets's wait and see ;-)
But; if we'd stay with mapnik we'll have to find out in the next few 
weeks/month, how we can perform with this setup. As already said this will 
strongly depend on how fast A. can get the renderer, but as what I heard in 
irc his first tests sounded pretty promising. But he probably won't be able 
to finish his work before end of next week. So I expect the next trial run 
earliest to start on the 5th of December.

> A lot of peoples requests to know at least the actual status of the project.

This is probably one of the most difficult question to answer. Sorry I can't 
really answer it.

> How will be long to render the entire world 

Currently it would take too long and/or we'd need too many machines and/or 
we'd need to much disk space.

> at each zoom level with the 
> actual code? 

if you want it up to zoom level 18; Very extremely too long. But as already 
stated speed improvement of the code is on it's way.

> Is there a batch queue running for the world? wath's it 
> rendering now? 

No. It's currently only run by half a dozen of individuals on there home PCs. 
But this is only kind of a test to see how it performs and how it looks .

> Has the foundation sufficient disk space to store *all* the 
> tiles? :)

This definitely is an interesting question. And browsing the wiki and 
mailinglist you can already see some calculations concerning this.

You see those are the question we even don't know yet by ourselves. But be 
sure we'll try to answer and solve them as soon as possible.

I hope these thoughts of mine are not too frustration or to euphoric. 

-- 
Jörg (Germany, Munich)

http://www.ostertag.name/
TeamSpeak2: ts2.ostertag.name, user: tweety, Channel: "GPS Drive"
irc://irc.oftc.net/#osm
Tel.: +49 89 420950304