Hi Andrea,
I am reading with interest this email, since I am nowadays extracting
all the JGrass processing power into a library that implements for
every module also the geotools process api.
My choice in the lib has been to rely on FeatureCollection and
GridCoverage2D, so I understand what you mean. Without the
subcollection methods I would be dead in some cases.

So I would be +1 to keep them and rely on FC and GC2D.

Ciao
Andrea



On Fri, Jun 11, 2010 at 3:09 PM, Andrea Aime <[email protected]> wrote:
> Hi,
> this week I've tried to make some of the processes in gt-process
> run in GeoServer, as they presented some decent variety in input
> and outputs. I succeded with some of them and found out a few
> issues in the processes that the attached patch fixes.
>
> The patch however does not contain only fixes but also some changes
> in the parameters.
> The problem I've found working on these processes is that they tend
> to have a large variety of input and output types, which makes it
> harder to deal with them from the p.o.v. of a fully automated
> client such as GeoServer is.
>
> In general, to allow a client to describe and handle all the inputs
> and output of the processes there has to be some agreement on what
> data types are supported and how to handle multiplicity.
>
> For example, the VectorToRaster has a java.awt.Dimension parameter.
> I don't believe it's something that should be part of the interface
> of a process interested in being used by a automated system, as
> it uses a rendering related class, and the same information can
> be easily expressed by using two doubles instead (which is the
> change I did in the patch).
>
> It would be nice to have a short list of types that it's safe to
> use in the in/out of a process that wants to be supported by
> an automated client (which could also be uDig, maybe with a generic
> GUI builder that uses the information in the input/output map
> to build the GUI to configure and run a process).
>
> What I have in mind so far are all the primitive wrappers, dates,
> strings, feature collection, grid coverage 2d, and referenced
> envelopes. Do you think the short list should contain something else?
>
> Another general question that came to mind was if processes should
> operate against feature collections or against feature sources.
> The issue with using collections is that it's harder to query them,
> in that light, removing the subcollection methods from the
> FeatureCollection would make coding efficient processes quite hard.
> Think of a process that needs to perform a vector overlay between
> two polygon layers, you scan one layer and look for polygons
> in the other layer that intersect it, in that case it's important
> to be able to rely on query methods.
> So moving forward I see two options:
> - the processes stop using feature collection and start using
>  FeatureSource instead
> - the processes use collection but we keep the sub-collection
>  query methods in the collection interface
>
> I was thinking more or less the same about coverages, instead of
> using GridCoverage2D we could trade coverage readers.
> However for coverages in theory we're playing with a set of
> JAI operations that get chained so, always in theory, trading
> coverage should be effective and more familiar to the programmer.
>
> Opinions?
>
> Cheers
> Andrea
>
> ------------------------------------------------------------------------------
> ThinkGeek and WIRED's GeekDad team up for the Ultimate
> GeekDad Father's Day Giveaway. ONE MASSIVE PRIZE to the
> lucky parental unit.  See the prize list and enter to win:
> http://p.sf.net/sfu/thinkgeek-promo
> _______________________________________________
> Geotools-devel mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/geotools-devel
>
>

------------------------------------------------------------------------------
ThinkGeek and WIRED's GeekDad team up for the Ultimate 
GeekDad Father's Day Giveaway. ONE MASSIVE PRIZE to the 
lucky parental unit.  See the prize list and enter to win: 
http://p.sf.net/sfu/thinkgeek-promo
_______________________________________________
Geotools-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/geotools-devel

Reply via email to