I thought it would be useful to summarize this. After any corrections to this initial attempt, I hope to put up a webpage for this.
We use the following Apache resources to store things: 1) SVN 2) p.a.o/www/uima.apache.org/d (part of our website) 3) repository.apache.org/ <- snapshot repo for uima 4) repository.apache.org/ <- "staging" repo for uima 5) maven central repo 6) p.a.o/www/www.apache.org/dist/uima The website is mostly stored in SVN, and is updated to p.a.o/www/uima.apache.org/pubsub and to the actual web server Apache uses, whenever there's a commit to ...uima/site/trunk/uima-website/docs. This use of SVN allows tracking of updates to the site. For larger generated objects, this is not an efficient use of SVN storage (my opinion), and we use the alternative of storing things to (1) above, directly. This is where our Javadocs (2.3.1 and later) go, without going thru SVN. (3) we use to store updates to the Docbook olink databases, as well as to store snapshot artifacts. (4) we use temporarily during a release - artifacts are released here, and then are voted on. Once the release passes, these are "released" to go up to Maven central repo (5). (5) This stores our Jars and other artifacts. The artifacts include a POM, and are signed and checksummed. The artifacts include the source (for jars) and for releases, you can find the combined sources (constituting our official release) and the zip version of the uimaj binary sdk (currently - used in building the uima-as distr). (6) stores the binary zip / tar artifacts, and our eclipse update site. This location is automatically "mirrored" by the Apache mirroring system. -Marshall
