Michael Baessler wrote:
Marshall Schor wrote:
Michael Baessler wrote:
Marshall Schor wrote:
...
Michael - when I looked into this, I think I found that to add a Pear to
a flow, you import the <pearSpecifier> descriptor. This is different
from the CustomResourceSpecifier. Is this correct?
Yes that is correct a pearSpecifier is similar to a
CustomResourceSpecifier but it is a separate independent descriptor.
I had assumed, from your question, that the CustomResourceSpecifier was
used when importing a pear, but I think this is not right. Please
confirm :-). Assuming I got that part right, does the CDE need to do
something about CustomResourceSpecifiers? Is this related to the the
Pear specifier somehow?
I think the CDE should be able two handle both specifier types.
When you say "handle" - do you mean that it should be possible to add to
an aggregate flow a descriptor whose type is
<customResourceSpecifier...>? What would be the meaning of this? I
don't think the framework would validate such a thing? Or did you mean
something different by "handle"?
-Marshall
To be honest, I currently don't know what the CDE checks if I add a
primitive ae to an aggregate.
There are 2 checks. One is for an acceptable descriptor type being
added to an aggregate. The 2nd check is that after the descriptor is
added, the underlying UIMA framework is called to "validate" the
resulting aggregate. I'm pretty sure this will fail if you add a custom
resource specifier to an aggregate - there's no support for that in UIMA
as far as I am aware.
Of course, if I'm mistaken, then we should add customResourceSpecifiers
to the list of allowed things in aggregates. But otherwise, I think we
should not do this, because one of the fundamental guiding principles of
the CDE design was to eliminate "errors" in building descriptors. If
you disagree, please give some reasons :-)
-Marshall
So what I mean with "handle" is, that the
CDE should be able to add a CustomResourceSpecifier or a pearSpecifer to
an aggregate ae.
-- Michael
