|
From: Martin D. <mar...@te...> - 2004-07-14 18:55:41
|
Hello all
I just updated geoapi-SNAPSHOT on iblio. The new GeoAPI JAR file
contains ISO 19115 interfaces, which can be browsed online here:
http://geoapi.sourceforge.net/snapshot/javadoc/index.html
Note that in order to compile Geotools with this new JAR, developper
will also need to do a SVN update.
This API is a realization of ISO 19115
(http://www.opengis.org/docs/01-111.pdf) as straightforward as I can.
There is no interfaces or methods addition compared to ISO 19115.
However, the API will probably change according GeoAPI mailing list
feedback. For example we plan to replace arrays by java.util.Set.
org.opengis.metadata.MetaData and Identification contains all methods
found in org.geotools.metadata.iso19115. Concequently, the later could
be removed. Key differences are:
- org.opengis.metadata has no MetaDataElement interface, because I
didn't found it in ISO 19115. However, we could add it if a
request is send to the GeoAPI mailing list.
- Methods in MetaData and Identification returns the more accurate
type instead of the generic MetaData[Element]. However, most
method names are the same.
This also bring an other issue. There is already an
org.geotools.metadata.MetaData class, which is a different beast than
org.opengis.metadata.MetaData. In the approach I took up to now, all
org.geotools.XXX classes are implementations of org.opengis.XXX
interfaces (i.e. I just replace "opengis" by "geotools" in the fully
qualified name). If we choose to continue that way,
org.geotools.metadata.MetaData should be an implementation of
org.opengis.metadata.MetaData. This means that current
org.geotools.metadata.MetaData should be moved in an other package.
Maybe some org.geotools.metadata subpackage? What do you think?
Martin.
|
|
From: Jesse E. <je...@re...> - 2004-07-14 19:08:42
|
Martin Desruisseaux said the following on 07/14/2004 11:53 AM: > Hello all > > I just updated geoapi-SNAPSHOT on iblio. The new GeoAPI JAR file > contains ISO 19115 interfaces, which can be browsed online here: > > > http://geoapi.sourceforge.net/snapshot/javadoc/index.html > > > Note that in order to compile Geotools with this new JAR, developper > will also need to do a SVN update. > > > > This API is a realization of ISO 19115 > (http://www.opengis.org/docs/01-111.pdf) as straightforward as I can. > There is no interfaces or methods addition compared to ISO 19115. > However, the API will probably change according GeoAPI mailing list > feedback. For example we plan to replace arrays by java.util.Set. > > org.opengis.metadata.MetaData and Identification contains all methods > found in org.geotools.metadata.iso19115. Concequently, the later could > be removed. Key differences are: > > - org.opengis.metadata has no MetaDataElement interface, because I > didn't found it in ISO 19115. However, we could add it if a > request is send to the GeoAPI mailing list. > > - Methods in MetaData and Identification returns the more accurate > type instead of the generic MetaData[Element]. However, most > method names are the same. > > > This also bring an other issue. There is already an > org.geotools.metadata.MetaData class, which is a different beast than > org.opengis.metadata.MetaData. In the approach I took up to now, all > org.geotools.XXX classes are implementations of org.opengis.XXX > interfaces (i.e. I just replace "opengis" by "geotools" in the fully > qualified name). If we choose to continue that way, > org.geotools.metadata.MetaData should be an implementation of > org.opengis.metadata.MetaData. This means that current > org.geotools.metadata.MetaData should be moved in an other package. > Maybe some org.geotools.metadata subpackage? What do you think? > > Martin. Hi Martin, Jody and I have been talking about the metadata and realized that there are a few things we need to talk about in order to make out metadata communication and work together. We're hopping on irc #geotools chat. It'd be great if you'd join us there. Jesse |
|
From: Jody G. <jga...@re...> - 2004-07-14 19:23:54
|
Jesse Eichar wrote: >> org.opengis.metadata.MetaData and Identification contains all methods >> found in org.geotools.metadata.iso19115. Concequently, the later >> could be removed. Key differences are: >> >> - org.opengis.metadata has no MetaDataElement interface, because I >> didn't found it in ISO 19115. However, we could add it if a >> request is send to the GeoAPI mailing list. > We needed to implement this in order to query against. (Cannot write a query api against bean properties, and still expect the same api to work against metadata xml documents). >> - Methods in MetaData and Identification returns the more accurate >> type instead of the generic MetaData[Element]. However, most >> method names are the same. > You may want to look at the use of reflection in the geotools Metadata class to see how both conditions can be met (easy java coding & metadata element for query). >> This also bring an other issue. There is already an >> org.geotools.metadata.MetaData class, which is a different beast than >> org.opengis.metadata.MetaData. In the approach I took up to now, all >> org.geotools.XXX classes are implementations of org.opengis.XXX >> interfaces (i.e. I just replace "opengis" by "geotools" in the fully >> qualified name). If we choose to continue that way, >> org.geotools.metadata.MetaData should be an implementation of >> org.opengis.metadata.MetaData. This means that current >> org.geotools.metadata.MetaData should be moved in an other package. >> Maybe some org.geotools.metadata subpackage? What do you think? > I think we are going to have more Metadata to implement then just ISO 19115 interfaces, dublin core was mentioned and for right now we need "off the cuff" metadata implementations for DataStore/GCE implementations of Catalog/CatalogEntry to work against. I was thinking of an abstract base class Metadata19115 implementing both interfaces that provides the required helper functions to make everything happy. Jody |
|
From: Martin D. <mar...@te...> - 2004-07-15 09:30:28
|
Jody Garnett a =E9crit : >>> org.opengis.metadata.MetaData and Identification contains all methods= =20 >>> found in org.geotools.metadata.iso19115. Concequently, the later=20 >>> could be removed. Key differences are: >>> >>> - org.opengis.metadata has no MetaDataElement interface, because I >>> didn't found it in ISO 19115. However, we could add it if a >>> request is send to the GeoAPI mailing list. >> >> > We needed to implement this in order to query against. (Cannot write a=20 > query api against bean properties, and still expect the same api to wor= k=20 > against metadata xml documents). Then, could you post a proposal on the geoapi mailing list please? Or=20 maybe we should do the IRC meeting first. > I think we are going to have more Metadata to implement then just ISO=20 > 19115 interfaces, dublin core was mentioned and for right now we need=20 > "off the cuff" metadata implementations for DataStore/GCE=20 > implementations of Catalog/CatalogEntry to work against. >=20 > I was thinking of an abstract base class Metadata19115 implementing bot= h=20 > interfaces that provides the required helper functions to make=20 > everything happy. It may not be easy. If I look at my recent experience, I tried to=20 refactor the legacy CS framework for implementing both the old API and=20 the new CRS API and quikly abandonned the idea: it was too messy.=20 Implementing two differents API in same time is very difficult for=20 non-trivial API. An other approach is to choose one API as the reference one (either ISO=20 19115 or Dublin core), and provide adapters from arbitrary metadata API=20 to the reference API. One advantage of this approach is to avoid the=20 introduction of yet an other metadata API. Martin. |
|
From: Jesse E. <je...@re...> - 2004-07-16 23:52:09
|
Geotools is still broken. A bunch of dependencies on Geoapi don't work. What we think happened was the ISO Metadata interfaces were changed but geotools were not updated. Or not committed. I'm afraid we're done for the day... I imagine it'll get resolved soon. Cheers, have a good weekend, Jesse |
|
From: Martin D. <mar...@te...> - 2004-07-15 08:15:47
|
Jesse Eichar a =E9crit : > Jody and I have been talking about the metadata and realized that there= =20 > are a few things we need to talk about in order to make out metadata=20 > communication and work together. We're hopping on irc #geotools chat. = =20 > It'd be great if you'd join us there. What about today (Thusday) 3 hours before the usual geotools IRC time?=20 It would be at 9:30 PST time if I'm calculating right. Martin. |