Current resource shortcomings

[MaximilianV]

So far, there are some flaws, imperfections or logical issues on the resource/dev branch, that need to be solved before merging:

• [x] The Rembrandt-Platform does not distinguish between Problem Definition and Optimization Method. But those have to be selected in the Gryphon modeler.
• [x] The resource validator currently relies on a hard-coded url of the resource manager, but should use a value in a config file. (There is a config file with this value, but it's located in public_src, not server_src.)
• [x] The resource validator relies on the resource manager mock and is not yet compatible with Rembrandt. This has several reasons: Rembrandt does not support input or output DOs coming from process instances at the moment. Rather, it relies on all resource instances available on the platform and optimizations are started without any inputs or outputs. This also implies, that Rembrandt does not provide incoming or outgoing resource types of optimizations which is a crucial information for the resource validator to work. (But the lack of these information is rather a fault of Rembrandt, not of Gryphon.)
• [ ] Gryphon requires explicit modeling of input and output data objects, even though they don't have any effect at deployment time. Neither are the modeled input data objects transferred to Rembrandt, nor does Rembrandt return the optimization result to write it back to the output data object.

[MaximilianV]

fa3aef846efc49c1ca1e337e175bc521efb2ace4 removed the Problem Definition from resource activities. Now only the Method can be selected, which is called Recipe in Rembrandt.

[MaximilianV]

In 52642e2b83e88d52b0b11c1c7ee21bcfc3b71fae, the hard-coded mock url was removed and replaced by a resource api helper. The api helper accesses the same config file as the gryphon front-end.

[MaximilianV]

c0a0ab4217b5c3288d0aacf796ff5c89adadb48f partly solves the third problem. Now the resource task validator relies on information coming directly from Rembrandt (including the inputs and outputs of a optimization method/recipe). But still, gryphon requires explicit modeling of input and output data objects, even though they don't have any effect at deployment time.