New Project -> Alexandria: Building all Apache Javadoc in one place (and CVSWeb interaction)

["Kevin A. Burton" <bu...@relativity.yi.org>]

Guys.

And now for my next trick :).  I have started a new project and I
thought it would be cool to let everyone know.

Basically I am going to take every Apache project done in Java and
provide a unified repository for all its Javadoc and generate it every
night.

This gives us the following features:

- only one place to go when you want access to CVSWeb
- generate Javadoc from all Apache projects
	- use "use" tag to see project interactions
	- generate all the javadoc from all projects so that we can link into
other projects
          without having to regenerate it's javadoc.
- support revisions so that we can migrate stable versions and not just
what is in CVS.
- Put the HTML behind a search engine.  This should allow us to share
code between project 
  without having to look through it by hand.
- Write our own Doclet to include the CVSWeb URL for obtaining the
source of what your are looking
  at without having to check it out.

If everyone agrees I would like to make this a dedicated Apache project
(not just something I run on my home system) under the APL.  This will
be a fairly resource intensive application considering all the
javadoc/etc.  Therefore it will probably need a dedicated home.  Anyone
want to volunteer?  It would probably need to crank for about 2 hours
with every project added.  The problem is that every night it would have
to regenerate all the javadoc since it doesn't support differential
builds.

Technically it is setup like this (I have some of this done locally):

- Defined a CVS Markup Language schema for marking up the CVS Roots to
use and their passwords, name, etc.
- XSL the CML into a navigation web interface so that you can easily
search through all the data.
- XSL the CML into a build.xml file under ANT.
	- this then does a "cvs update" and a "javadoc" target for all the
roots
- After that we just need to tell ht://dig to update it's index

This also gives us a chance to really eat our own dogfood.  (Xerxes,
Xalan, Cocoon, Ant, Apache)

Why the name Alexandria?  Alexandria was this huge Library in ancient
greece that was destroyed by fire.  May historians consider this a huge
loss of knowledge.  I just thought it was appropriate considering we
have all this structured information under Apache and we haven't used
it.

-- 

Kevin A Burton
http://relativity.yi.org
Message to SUN:  "Open Source Java!"
"For evil to win is for good men to do nothing."

Re: New Project -> Alexandria: Building all Apache Javadoc in one place (and CVSWeb interaction)

[James Duncan Davidson <ja...@eng.sun.com>]

> Basically I am going to take every Apache project done in Java and
> provide a unified repository for all its Javadoc and generate it every
> night.

Kick ass. You're right, this may be too resource intensive for locus to take
on considering that it's going to be long run times, but let's see what we
can do to find a home that you can run on.

.duncan