On 12/01/12 20:13, Robert Vesse wrote:
We actually reference it as a library (via maven) and call
Fuseki.init() in our startup code to actually start up Fuseki.
We then have our own init() routine which registers all the relevant
query engines and does other setup - custom describe handler,
optimization config etc
Even if we were using 0.2.1 I think we would prefer not to use
assembler descriptions as that would allow users to break things by
changing the file. Also we configure differently depending on the
command line arguments our code received and I don't know that
assembler would allow us to cope with that nicely (though I think we
can call arbitrary code?)
Rob
Rob,
Thanks for the background. The way a server is created programmatically
has changed but not so as to block what you're doing. Essentially, the
configuration file is parsed to a set of dataset descriptions and the
set of descriptors is passed to server creation. There's no reason code
couldn't create such a descriptor. Learning from Joseki, I'm aiming for:
configuration -> java data objects -> build server
The data objects are structs - simply recording the configuration - and
can be built by code as well.
Andy
