On 2010-09-21, Philip Gillißen wrote:
> Ok, this may be an infrastructural problem that cannot be solved with a
> "tah2boinc" scriptlet. It looks like there's more to fix, doesn't it?

It has been a while since I last ran boinc, but I am not sure it helps
us a lot. AFAIK, it is good at dividing workload into batches that
clients fetch from time to time and crunch it, and then upload the
results from time to time. t...@h on the other hands wants to immediately
hand out a new request as it is made and as soon as it is rendered we
want to send the result to the server for quick turnaround times. Not
sure if boinc could fit that bill.

Also afaik boinc clients bring all they need with them in a local client
installation, not sure if we can put inkscape and perl (with all needed
modules) and all the other stuff in a local client package. At the very
least it would require coding and maintaining a boinc server /
infrastructure. And someone would need to code a boinc client and
integrate it with the t...@h client stuff. We have barely enough (programmer) 
resources
to just keep the t...@h client running, so that is why you don't see
anything happen, even if everyone thought that this is a great idea.

I for my part keep the server running and extend it as needed, but I am
not very motivated to set up a boinc server infrastructure that I don't
know anything about.
 
> All this consideration lead me to one question: Is TaH dead?

This is a question that has been asked regularly, and my reply to that
is: we have 218 user accounts rendering in the past months, and we are
keeping a tile server running that can be accessed with no restrictive
terms of service (and no uptime guarantee :-)) which is separate from
any for-profit entity and from the OSMF. We have people experimenting
with style sheets in the t...@h layer.
Are lots of exciting development things going on in t...@h world? Nah, but
we are keeping it up and running. Is that dead?

> It looks like the "at Home" addition is lost and does not create any
> advantage.

It is not the most efficient way of rendering the world, but it allows
the server to be a rather modest thing (the t...@h server could not cope
with a postgres/mapnik installation I guess, not sure what the
requirements are). All the hard stuff is distributed to the clients and
whenever tile.osm.org fails, we still provide up-to-date tiles.

> A new infrastructure for "TaH v2" would be necessary. Is it too radical?

We are all ears for improvement suggestions. We just need the people who
also code them :).

Sebastian

Attachment: pgp3vaqqYzQfl.pgp
Description: PGP signature

_______________________________________________
Tilesathome mailing list
[email protected]
http://lists.openstreetmap.org/listinfo/tilesathome

Reply via email to