NOTE: THIS IS NO LONGER NECESSARY AS OF TILESTACHE 1.6:

http://tilestache.org/doc/#wsgi

gunistache is a very simple CLI wrapper to run a WSGI-derived TileStache server on top of gunicorn.

It exists only because there is still no way to pass an application-specific config file (that is something that your code can read at runtime) to gunicorn.

It works likes this:

$> /gunistache -w 4 -k egg:gunicorn#gevent -b localhost:8000 config.json

What happens under the hood is this:

• The example_config.json file is parsed and checked for a gunistache['name]' key which is used to generate a new gunicorn script/handler with code to render and cache map tiles using TileStache.

• The example_config.json file is the same config file/data that TileStache uses to know what types of tiles to generate and how to render them. The path to the config file is hardcoded in the new gunicorn script/handler.

• By default these new handlers are written to a directory called 'gunistache' in your operating system's "tmp" directory. You can override this setting by specifying a gunistache['app_dir'] value in the config file. For example:

  "gunistache": { "name": "tileserver", "app_dir": "." }

• Once that is done the code is executed by handing everything back off to gunicorn.

That's it. There are no ponies. Some day this code will (hopefully) be unnecessary.

See also:

• http://github.com/benoitc/gunicorn

• http://github.com/migurski/TileStache