From: Ian S. <Ian...@ar...> - 2003-07-17 07:32:33
|
Hi, Just finished the mind numbing task of merging the features-exp branch onto the trunk. The existing code was tagged as pre-features-exp-merge. There are some build problems, but mostly in very good shape. modules with problems: ---------------------------------------------------------------------------------------------------- lite-rendering -> see Java2DRendering wmsserver -> relies on lite-rendering + some core/defaultcore problems gui -> problems, relies on lots old core/defaultcore API OcacleTest has stange formatting (its fixed build wise- oh yeah, now I remember, I saw a bunch of ^M thingees)? pickle -> I'll get to it tomorrow, unless someone else really wants to do it svgsupport -> couple of minor EnvelopeExtent (old API) refs + 1 more algorithms -> a bunch of old FeatureType api usage + more gtopo30 -> a few minor old feature api usage For examples as how to fix the various errors, consult similar code in the modules that build. defaultcore is probably the best place to look... Some observations made whilst running over various code... Notes: -I realized the introduction of the NULL_ATTRIBUTE object causes problems when the client code is casting. For instance: String s = (String) feature.getAttribute(...); WIll cause a problem... The objective was to remove the ambiguity when looking for an attributes value in a given Feature, that is, is it null, or does it not exist? Before an exception was thrown. I say, its up to the caller to interperet the return value. Is it null cause its not there (maybe important, maybe not) or is it null cause its null (maybe important, maybe not). The actual presence of the AttributeType can be determined if its neccessary. Any thought? -move AbstractDataSource to core (otherwise all datasources depend on defaultcore) Good night, Ian |