Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
[wtp-releng] Projects need to handle their own JavaDoc generation, bundles and features


There's one item of business that has almost fallen through the cracks, but ...

I've been "warning" for some time, in the weekly status meetings, that projects would need to generate their own JavaDoc and corresponding bundles. But no action so far.
The current and probably even previous ones are broken see bugs 225914, 210844, and the "WST and JST" distinction is no longer meaningful and even confusing, some of the overview documents are now out of date and incorrect (with no one "owning" them any longer), and our releng team (i.e. me!) can not provide this as a "central service".

So, I'll be officially removing the (broken) wst.doc.isv and jst.doc.isv bundles and features this week.

I am sorry if this seems "sudden" (and it is) but I've been hoping the Server Team would provide an example solution, but that doesn't seem possible after all, so I do not know how else to solve the problem, except to recognize that it is the sub-project's problem, and their responsibility to solve.

Many of the sub-projects do already do this (JSF and JPA) and even some of the components (Facets and _javascript_, I think). So ... should be possible. Only problem is that no one has some up with a generic "fits all", mostly automatic way of doing it. I believe at least some of the projects use the tools in the PDE to generate their JavaDoc, and then simply check it in to CVS, all ready to go, so to speak. (Only issue with this approach, is those teams must remember to re-run their workspace functions occasionally to capture their JavaDoc edits).  Another approach some of you might enjoy is to take the current wst.isv.doc or jst.isv.doc bundles and hack them up to do what ever you'd like. (Just be careful with this approach that it works in a batch build, before you rely on it ... JavaDoc generation can take a huge amount of memory while running) and often results in "out of memory" errors.

I thought the convention was to use "doc.isv" for such bundles, but doesn't seem like everyone has followed that ... so, not sure what conventions you all are using.

So, I'm sure questions and suggestions (and communication, in general) on this list are welcome and collectively all you project leads can solve this problem.

Thanks,


Back to the top