Hi Greg, >>- a Color Utility class to transform/check/simulate them , as >>discussed via mail previous days ... > Sounds like it could be valuable. When you have something ready, why don't > you send it around for review. OK, but some time is needed, i think this have a low priority ...
> Ha. I've actually tried to implement such a serializer several times, and > have come to the conclusion that there is no such thing as a "simple XML > serializer". WTKX is the closest I could get to it. No, i was thinking at something simple and not necessarily general, like a Pivot convention to simply import/export data in a customized xml format, like the JSONSerializer does (but in this case it's using a standard and more simpler format). And could work as the JSONSerializer, handling only some types of Pivot Collections and provide an out custom format. This is only an idea, maybe could be too much complex for a little general usage ... I'll try to make some experiment on this. Otherwise, for a more general solution, some time ago I've seen something using XStream, to simplify many of these aspects ... so maybe we could think at a generic XMLSerializer to pass a generic implementation (Stax, XStream, etc) to do the real code/decode work. In any case: low priority, ok ? >>- move projects layout to a more common standard, like that used in >>Maven for libraries (jar) projects, so for example throw -test >>projects moving their code inside the related project, in the test >>directory ... ok this maybe for 2.x > > So, we'd have a test subdirectory under each project that would contain > source code? Yes, for example the Maven standard layout for a library (jar) project is like this: http://maven.apache.org/guides/introduction/introduction-to-the-standard-directory-layout.html Then, for example we could thing at using some of these dir, like src/main/resources to put inside wtk files (if useful) and other resources, and add some others like src/test/webapp to put test web applications needed to publish data to pivot clients, etc ... Wicket for example is aligned with this, and many other frameworks, also that not using maven, so why not do it ? Oh, sorry, one little thing of Pivot source projects: will it be possible to rename them, having a name like pivot-core, pivot-starter, etc ? Having all them in a workspace mixed with others projects can give confusion ... > Maven support is being actively discussed. Great, Ant for most common/simple uses is enough, for mid/long term adding also maven support could help, and also to integrate Pivot in mid/high complex Java EE projects. > I'm not sure about an "experimental" project, but I would like to discuss > alternatives to our current approach to handling external dependencies (i.e. > libraries that don't come with the JDK). Currently, we store them in SVN > under in project-specific /lib directories - is there a better way? Maven handle this without including dependent jars (and related dependencies), because copying other jars in our lib dir could give us some licensing problems i think (depending on the license type), but on this our friends from Apache could help us better ... But a problem i found is when some jars aren't in Maven Repository (like many jars from Sun), and in this case any developer needs to manually install in his repository, but with some command all works. If useful, i could post my preliminary version (I'm not a maven expert, so any help on this is appreciated) of maven pom.xml file for the Pivot Starter project I've experimented some months ago (useful to do builds, run standard and test executions, checks on code, etc), and if wanted the full zip (only few KB), but the best could be at least to have the right paths inside, so a little update of URLs inside should be done before this. Tell me. Thanks and good work, Sandro
