it's command line, windows (dos) ... simple for a non-programmer like
me to understand.
... the user just needs to set the co-ords. :)
Now that i think of it ... it shouldn't be that hard to just make a
full planet script to get the 1x1 tiles :)
... but for specific cities, manually geeting a box is needed...
lambertus did the heavy work of figuring out nice boxes anyway..
cheers,
Sam
--
Twitter: @Acrosscanada
Blogs: http://acrosscanadatrails.posterous.com/
http://Acrosscanadatrails.blogspot.com
Facebook: http://www.facebook.com/sam.vekemans
Skype: samvekemans
IRC: irc://irc.oftc.net #osm-ca Canadian OSM channel (an open chat room)
@Acrosscanadatrails
So with reference this this diagram, FOSM is currently set up with the
PostgreSQL Backend supplying API 0.6? I am assuming that planet
dump/diffs will be provided, and then we just need to set up a server
to fetch these dumps/diffs to do what is in the Mapnik box on the
diagram (load into postgis, render tiles using Mapnik with mod_tile
and serve the tiles)? This shouldn't be too hard. I haven't looked
into mod_tile before, so I'll have a look into in.
Also where does the rails port fit in with this, and is it set up on fosm?
If that is some GNU/Linux server, I think I could.
(I hope it might be Debian Squeeze or something similar as I'm most familiar
with those; while I should be able to manage other variants too, it might be
quite slower / more time demanding...)
Quick proof of concept (just Croatian map on slow server with not much
bandwidth nor disk space) can be seen at:
http://osm-mapnik.torres.voyager.hr/medium_my.html
The slippymap interface could be more like OSMs of course, this was just
quickest one to implement to show it works...
(note: it uses mod_tile+renderd to render on demand [as the osm.org site
does], so if you get broken image links just wait dozen seconds and
zoom out/zoom in, and they should be there, rendered just for you)
> Please get in touch if you can help.
Contact me directly on e-mail (as I read it more often then lists) if you
haven't found better volunteer yet.
> And if you don't know anything about map rendering but would still like to
> help in some way then please also get in touch, there are many tasks that
> need to be done over the next three months.
I'd like to help in setting up / sysadmining tasks as my free time permits
(which is unfortunately variable and prone to run out when it is most
inconvenient, mind you...)
--
Opinions above are GNU-copylefted.
I repeat my offer to implement it on fosm.org if it is wanted...
> So, what is the status of browsable FOSM map?
>
> I repeat my offer to implement it on fosm.org if it is wanted...
>
Yes, some news on this would be appreciated
Liz
> Sorry folks, I've been very busy with other stuff for the past couple
> of months.
>
> As soon as I get enough time I'll be setting up a Debian based server
> which can be used as the rendering platform.
>
> Matija, I would like to accept you offer of help and will contact you
> when I begin to setup the server.
>
> 80n
thanks for the update 80n
Great, whenever you're ready.
Do you perhaps have any estimate when that might be (approximately)?
I ask because it would help me organize time (so that I do not accidentally
dive into some other time consuming tasks or similar exactly at the moment
when I will be needed for FOSM).
Not sure if it's possable to get a hikebikemap rendering system going,
then the code could just be copied.
If 80n needs the 1st mapnik rendering to be created, then i guess that
should be a priority.
I have a feeling that NZ would end-up having their own server API
setup. ... is it easier to setup an independent regional API, then
just have edits syncing back to fosm.org?
... would that lighten the load?
... and perhaps the map could render with a delayed reaction?
Is this possable?
cheers,
sam
--
---
Across Canada Trails - Beyond 2017 - The National Trails Network
Victoria, BC Canada
Twitter: @Acrosscanada
Blog: http://acrosscanadatrails.posterous.com/
Facebook: http://www.facebook.com/sam.vekemans
Skype: 'Sam Vekemans'
Member, CommonMap Inc. http://commonmap.org/
IRC: irc://irc.oftc.net #CommonMap
Also find us on Facebook, Twitter and LinkedIn
Great, whenever you're ready.
Do you perhaps have any estimate when that might be (approximately)?
I have a feeling that NZ would end-up having their own server API
setup. ... is it easier to setup an independent regional API, then
just have edits syncing back to fosm.org?
... would that lighten the load?
... and perhaps the map could render with a delayed reaction?
Is this possable?
If someone has a spare harddisk I could install it in my server and it
could do as a temporary tile.fosm.org although the link is not super
speedy. I'd then just need your public key and desired login and you
could try setting it up. Right now I'm hesitant to let postgres and
tilecache on the same disk with the rest of my data, but the machine
has relatively good computing power.
Cheers
Do you want money for a drive, or actual hardware?
I would have thought it'd be cheaper to buy hardware locally than have
it shipped internationally...
Yeah, probably. But either will work and I can later pass the disk to
someone with a better connection/hardware when it's available. (it
could be a laptop drive or a microdrive / CF card too -- I don't know
what their usual performance is though.. I think CF cards should be
fast?)
Cheers
CF is relatively slow and low capacity compared to SSD (Solid State
Drives), the downside of SSDs is high failure rate, Seagate put it at
above 30%...
A suitable rack case of 12+ HDDs in the right configuration offers a
lot of IO per second...
--
One problem I foresee if you'll still need some way to keep rendered
tiles up to date, as I'm assuming archive.org only does hosting,
alternatively it might be a good idea to see if they can spare some
CPU cycles to render tiles as well as the data is updated.
This sounds like a quite bad idea. Map tiles are a product of
rendering data, they're like binaries, not source code. Versioning
autogenerated content is usually avoided but in this case it's a
necessity. You'd quickly fill up even the most generous host trying
to keep subsequent versions of tiles for any reasonably sized area. I
also think archive.org is doing a great service to the society and
it'd be a shame to waste their resources for keeping autogenerated
content, especially autogenerated and quickly becoming outdated
content (an OSM tileserver can generate gigabytes of new data every
minute), you know it kills hard drives quickly. Also there's a reason
modtile renders on demand.
Cheers
Some times I wished in the past to see previous version of tiles to
see what has changed, yes it'd be a massive waste of space, but it'd
still be nice.
Perhaps instead of focusing on rendering server side we should be
focusing on rendering client side.
you can purge old versions in git as well, even on the server
http://www.kernel.org/pub/software/scm/git/docs/git-filter-branch.html
see also
http://stackoverflow.com/questions/930612/git-remove-oldest-revisions-of-a-file
and with submodules you can reference other repositories and select
versions, that can be used to weave many repos together.
> also think archive.org is doing a great service to the society and
> it'd be a shame to waste their resources for keeping autogenerated
> content, especially autogenerated and quickly becoming outdated
> content (an OSM tileserver can generate gigabytes of new data every
> minute), you know it kills hard drives quickly. Also there's a reason
> modtile renders on demand.
WIth archive org we can replace files and keep only the new ones.
>
> Cheers
> Perhaps instead of focusing on rendering server side we should be
> focusing on rendering client side.
yes, I think we should work on client side rendering with many many
people helping , tiles at home etc.
I think we could even render only the changed layers when updating the
repository.
we can also run mapnik at home, I am experimenting with that.