[ZOO-Discuss] ZOO Execute requests

[Soeren Gebbert]

Hello,
more below:

2010/5/17 Gérald Fenoy <gerald.fenoy at geolabs.fr>:
> Daisuke-san,
> I really thank you for this effort as we heard a lot about the needs for
> this kind of virtual machine image.
> About the error messages you get in your apache error log, it is only a
> debug message displaying what was passed by ZOO Kernel to the Python module
> function as output value. Indeed in the released ZOO Kernel it now fill the
> output dictionary with the default output defined in the zcfg file if none
> was specified when the service was requested. This way you can overwrite the
> "value" field only from your Python module rather than creating a full
> output dictionary. I think it is better to handle things this way.
> A note here about default values, in the released ZOO Kernel we also fill
> all the default inputs parameters even if they are optional. I would like to
> get feedbacks for this specific implementation. Indeed, in some process it
> makes sense to handle things this way, the Buffer service for instance get
> then the default value for BufferDistance input but I wonder if it makes
> sense for each processes. Indeed, some process, like the ones shown by
> Soeren using GRASS a backend, seems to define some parameters with default
> value but don't want to get them filled to ensure that they wasn't passed to
> the service. In fact in this case they are really optional. In the Buffer
> service, even if the BufferDistance is optional, then in the code of the
> Buffer function we have to use a default value. So I thought that it can be
> a good thing to get the default value directly filled by ZOO Kernel, this
> way we don't have to think about the default value (as they are settled in
> different files, the defalut value was set in the zcfg and the default value
> should also be used in the code of the service function). So if the code
> evolve the zcfg have to be corrected to follow the new implementation.

I have just created a ticket about this topic in the ZOO bug tracker:
http://www.zoo-project.org/trac/ticket/3

The problem with optional literal data inputs and default values is,
that they are also filled by zoo kernel when no default value is
available (i.e: BufferDistance=NULL). In case of grass this results in
an error. So we may have to rethink the concept of filling optional
literal data inputs?

Optional complex Inputs should IMHO not be filled by default.

Best regards
Soeren