Hi all, I'll need to cut own my efforts in [EMAIL PROTECTED] programminga bit, now that things seem to run smoothly, I do have a regular job...
However, the server records the client id's that Matthias Julius has recently implemented, you'll see that the log file states now things like: 08-08-11-00:39:06 DEBUG Saving tileset at (maplint,12,784,1645) from user matthias_julius_net_net (client id 9555) For now they are recorded in the log file, so in case of a failed client we can go and look up the client id. I have modified the client to generate random client uuids between 0-65535. In case your client still has a bigger number, the server will just take the last four digits. As a uuid (slightly misleading name) only distinguishes all clients of a single user, there's now need for astronomically big random identifiers. I plan to also store the client_uuid in the "request database" which stores entries for 48hours after uploading a tileset, so that I can display the "last uploads" of each user together with the corresponding client_uuid. But this is not extremly urgent on my list. I think most improvements can now be made on the client-side. To my mind comes: 1) assemble tilesets on the clientside already, don't upload zip files 2) check last-modified age of tileset, compare with API age and skip request if the tileset on the server is younger already 3) return a request if the tileset complexity is higher than what the client is willing to handle 4) Fork the rendering in 2 threads, have one already downloading the next task, while the first is still rendering the tiles. (downloading can take quite a while, but doesn't consume lots of resources - at least on the client side, so we could speed up rendering by doing this already) spaetz _______________________________________________ Tilesathome mailing list [email protected] http://lists.openstreetmap.org/listinfo/tilesathome
