Re: [Geotools-devel] Geotools and GeoApi

[Martin D. <mar...@te...>]

Hello

Harald Wehr a =E9crit :
> thanks for your answer. I would like to commit something to this work.=20
> Are there already some papers/plans about the steps that have to be=20
> done? It seems to me that this is a fundamental change of the geotools=20
> basis isn't it?

Yes, it will be a major change for the whole Geotools library (should we=20
call it Geotools 3? :)

As I see it, the plan are:

1) Move to Subversion on JIRA as soon as we can (such a major change
    will be much easier with subversion than CVS).

2) Merge modules as suggested in the IRC one weeks ago. Again, with
    such a change with repercussion everywhere in Geotools, I think it
    will be easier if we have one core instead of a bunch of separated
    modules that should be related.

3) Refactor CTS first (my estimation: 1 month).

4) Refactor GCS next (my estimation: one week; this one has no major
    change and should be straightforward).

5) Write adapters wrapping JTS objects into ISO 19107 objects.

6) Write adapters wrapping Java2D objects (e.g. GeneralPath)
    into 19107 objects.

7) Change every JTS dependencies into ISO 19107 dependencies.

8) Refactor j2d renderer. My estimation: at least 4 months. It will
    be a big and long task in order to bring the renderer to GO-1. I
    will try to fix most JIRA bugs reported against the renderer in
    the process.


What would you like to commit? Someone need to do #1 and #2 (probably=20
James?). For myself, I plan to do #3, 4 and 7. We need someone else for=20
#5, 6 and 7. Actually, the work on #5 and #6 could begin right now.





> Do you think that some adapter classes would be sufficient here? I=20
> thought that the JTS packages is made for analysis of Simple Features.=20
> In 19107 you have a lot of further geometry classes like arcs and so on.

No, JTS will not cover fully ISO 19107, but could be a nice starting=20
point. JTS will be one implementation of ISO 19107, not *the*=20
implementation. The user would be free to chose between a JTS or a=20
Java2D backed implementation, and we may provide more elaborated=20
implementations in the future.


	Martin.