Adrian Custer was kind enough to ambush me on IRC for a bit of planning 
.... here is a summary.

1) I'd like to push us to graduate by the end of the month

It sounds like we can do a couple of things:
1. search and replace to update the (c) headers
2. chew through the code review for most of the core libraries

After that we have an open question; what to do with module that are not 
ready?

For this to happen we need to have the following information filled in 
before we annoy the OSGeo board:
- http://wiki.osgeo.org/wiki/GeoTools_Incubation_Progress
- http://wiki.osgeo.org/wiki/GeoTools_Provenance_Review

We should probably move these pages to our own wiki in order to just get 
things done...


2) wha'ts the status of main's dependency on epsg-wkt?

It is just a test dependency; main should not depend on epsg support ... 
it is CRS Authority agnostic
<dependency>
   <groupId>org.geotools</groupId>
   <artifactId>gt-epsg-wkt</artifactId>
   <version>${project.version}</version>
   <scope>test</scope>
</dependency>

3) in passing, I'm going to have to do udig svncleanup on my own time, 
maybe this weekend

Go acuster go ...

4) what do I do with iso geom module?

Talk to gdavis (the module maintainer) and see if acuster can take 
over.  There are plenty of problems to fix, the goal is to keep the test 
cases (and test coverage high) and drop the module back to unsupported 
for refactoring.

I personally hope that the GeoAPI GeometryFactory constellation is 
reviewed as well.

5. A. pull gt.util out of metadata and gt.measure out of referencing to 
make a gt-base module

This is an interesting one, on a related note I have considered killing 
gt-api module because it prevents user from just using the layers of 
abstraction they care about (ie they drag in gt-api for the JTS utility 
class, allowing them to use JTS Geometry + CRS) and they see stuff like 
DataStoreFinder...

Diagram here: 
http://docs.codehaus.org/display/GEOTDOC/02+Meet+the+GeoTools+Library

One suggestion is to set up gt-util as the public API for metadata and 
referencing; forcing the JTS dependencie as provided so it is useful to 
the real world. And keep gt-api as the api for coverage and higher.

Next suggestion is to make it called gt-geotools and use it to contain 
the GeoTools utility class and our Factory system etc... and dissolve 
the gt-api class ..

This one needs further thought; basically we need to think about the 
user community on this one

6) I've tried to digest your vision for expression as a mechanism for 
pulling geometries out of features and every time I get stuck

Q: who does the evaluation of the expression?
A: Who cares; I need you to let that go for a bit until you see what 
Expression and Filter are for.

Q: where how?
The user is letting us know what they want through a data structure

Q: what is the fallback
A: The fallback did not exist; until the SE 1.1. specification added it 
- and even so it is only there as a fallback for when the "function" the 
user asked for is not avaialble. Other fallbacks exist in the form of 
NULL checks and existence checks but those are up to the user to use.

Q: I really just want to know what we have to write to pull geometries 
out of features?
A (as a user) You need to write an expression saying where the geometry 
is (ie a propertyName) or an expression to calculate the geometry; 
perhaps a function?

A (as a developer) You need to write a Converter; the implementation of 
Expression uses our Converters API to handle the requirement that 
Expressions not care about type...

Q: so who is smart? the user code? the feature? the data store? ...

It is the DataStore's responsibility to be smart; they are the code at 
the end of the day who is going to look at what the user wanted and do 
its best to figure it the heck out.

We have provided a lot of tools to help them be smart. In the form of 
FilterVisitors that prepair the expression (perhaps applying 
environmental vairables like DPI). In the form of the 
Expression.evaulate methods itself; which delegates to the Conveters API 
in order to take care of all the "type" information that the user is 
supposed to be able to ignore.










-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft 
Defy all challenges. Microsoft(R) Visual Studio 2008. 
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Geotools-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/geotools-devel

Reply via email to