On Friday 18 April 2008 01:07:29 am Jody Garnett wrote: > Gabriel Roldán wrote: > > Truth is, I'm afraid we don't have enough modeling power as to make an > > AttributeType bound to a List and still know a list of what... > > Let me try; that is an AttributeDescriptor with minOccurs=0, > maxOccurs=Integer.MAX so you would model a gml:posList element like that? leading to potentially thousands of Attribute instances, one for each Double instance? And also does it really capture the semantics of a List? (we only have the notion of a Collection afaik) > > > I seem to remember we discused about that a couple times and never got to > > an agreement. Some told it should just be a ComplexAttribute bound to, > > eg. Int, and multiplicity 0..*, others it should be its own > > ListAttributeType interface, keep being simple but extended to expose its > > content type (int, long, etc). All the grief due to generics runtime > > erasure? > > The problem is indeed split into two: > - AttributeType captures the description of what is being used (like a > Java Class does for programming) > - AttributeDescritptor captures the part that describes how it is used > right here (like a Java field does for programming). How it is being > used covers nillability, minOccurs, maxOccurs etc... > > > Which recalls me we still also don't have a clear mapping of xml > > attributes to our FeatureModel, and are making use of tricks like > > userData but without a well known key (ie, > > putUserData(javax.xml.sax.Attributes.class, Map<String, String>) and > > monstruosities like that) > > I don't think userData tricks were needed at the end of the day... > > > These are basically the two gaps I still see in the FM once we got rid of > > "xml particles" like choice, sequence, etc for good. > > The xml specific subclasses were removed since they were only holding > javadocs and did not contribute any API. > - ComplexAttribute defined as a List<AttributeDescriptor> is a good > mapping for sequence > - ComplexAttribute defined as a List<AttributeDescriptor> where all are > optional, and minOccurs=1, maxOccurs=2 is a good mapping for Choice > - finally ComplexAttribute defined as a Set<AttributeDescriptor> covers > the last xml use case yup, that's what I tried to mean by "for good".
Gabriel > > > Sorry Anaxa I'm not answering your question... guess there's not a clear > > answer actually, but place to debate. > > Cheers, > Jody > > !DSPAM:4045,4807d83a185922458217002! ------------------------------------------------------------------------- This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone _______________________________________________ Geotools-gt2-users mailing list Geotools-gt2-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/geotools-gt2-users
