[Tilesathome] Purpose of Tiles at home

[Matthias Julius]

Sebastian Spaeth <Sebastian at SSpaeth.de> writes:

> Hans F. Nordhaug wrote:
>> Me: I have just installed the Tiles at home client on my server - nice to be
>> able to contribute some CPU cycles to the OpenStreetMap project.
>> 
>> Shaun McDonald: um do you know that one machine can deal with the
>> rendering of the whole world wilst T at H needs lots of machines and
>> bandwidth?
>
> Well, that is their view and they are certainly entitled to it. Others
> see it differently. tiles.openstreetmap.org will cut your IP if you suck
> too many tiles and there was some discussion in the past as to enforce
> stricter limits. That is no criticizm of the foundation, they probably
> have to do that in order to keep their bandwidth under control and of
> course there is also a commercial startup that would like to sell these
> services. tiles at home on the other hand has so far never ever cut of a
> single IP for downloading too many tiles.
>
> Also see Freeriks response: tiles at home has provided close-to-realtime
> map updates since hwow many years? At least March 2007. It's only a
> rather short time ago that mapnik started showing hourly (and minutely
> now?) updates.

And here the question arises whether this has made T at H obsolete.

> There are also some differences in style creation. Mapnik is centrally
> driven, while t at h is much more open to community "enhancements"
> (sometimes an advantage and sometimes not).

One could follow the same approach when maintaining mapnik styles.

>
> So, given the historical reason, I wouldn't say that it's a worthless
> piece. Yes, it is more effort than a single mapnik instance, but it
> runs, it is independent from the foundation and any commercial interests
> and the styles are very much user-driven.

I don't think it is justified to keep T at H running just for hysterical
raisins.  If mapnik is so much more efficient one could replace the
T at H render infrastructure with a mapnik server using the same styles.
Users of the map probably won't see the difference.

>
> Will it hurt if we turn off t at h at some time in the future? Probably not
> that much, but it already has had its benefits.

The subject can be devided into two questions:

- Does anyone care for the map layers T at H produces?

- Is the way the layers are produced obsolete?

One strength of T at H is its modular renderer.  It is relatively easy to
implement another preprocessor to do some fancy rendering.  How
feasible would it be to have mapnik render the maplint layer?

Another point for T at H is the easy way to setup another render
instance.  How do you do that with mapnik if you don't have a potent
server available?

Yet another argument for T at H is that it seems to work pretty well
since quite some time now.  (I am not saying that mapnik doesn't.)
Why change a running system?  It does not seem to require a huge
amount of maintenance.  When people loose interest they will start
contributing and the tiles will become horribly outdated.  Until then
I would not worry about purpose.

Matthias