geronimo-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "n. alex rupp" <>
Subject Re: console-web (notes on collaboration between non-committers)
Date Mon, 03 Nov 2003 19:06:52 GMT

If you'd like, I can grant you commit access to the service-cache project at
sourceforge so you can work with a net while building your console.  I started
that "project" so we could work together and prepare potential submissions for
Geronimo and benefit from access to CVS.  Matt K's working on a CLI management
interface there (last I heard--I don't see a lot of activity).  Otherwise, it
might suit you better to host your own project CVS.  I'm a big advocate of the
DIY route--to take your future into your own hands and administer your own CVS
tree is very liberating.  I recommend everyone tries it.

In any case, the CVS tree at Geronimo is not a good place for experimental work
or speculative R&D between non-committers and especially for those of us not
working within the context of a standard.  It's a good place for tested,
functional code and directed efforts.  This is why I started the service-cache
to begin with, because the project committers shouldn't be expected to maintain
my experimental code at a fine-grained level.  They don't need the overhead and
I don't need the friction.

Geronimo is just too modular in design for us to limit ourselves with the notion
that we need to do all of the work *here*.  I'm glad to have all of the
conversations here, as it *is* the list of record, but we can write our own
pluggable modules (how else could you describe a web application?) and maintain
them elsewhere.  Eventually, we'll be able to register them with Ibiblio and the
server should be able to download and install them automatically at runtime.  If
the PMC and the committers want to make our modules the default functionality
for the Geronimo stack, so be it.  I'm happy to structure the work so it is
easily injected into Apache's CVS tree, happy to build the diffs, submit them to
the JIRA, and happy to work closely with the committers in supporting the
product--but I'm not dead in the water if they're all busy, or out of town, or
at ApacheCon.  That's the beauty of the DIY approach.

I know this management application is going to need several rewrites at some
point down the line, but I'm trying to take the agile route and to focus on
implementing features the developers need *right now*, like object invocations
(which I'm almost done with, screen shot attached).  I've gotten trapped in the
past by overengineering in the early phases of development.  It stalls visible
progress and can be quite discouraging and frustrating, plus you never really
understand the needs of an application until you've already built it incorrectly
a few times (ask Jeremy and Dain about this).  And those needs change, so it's
ridiculous to plan too far ahead.

Finally, as far as I'm concerned, the two most important features of this web
application are that it's easy to use and that it pushes the aesthetic limits of
the medium.  And yes, I'm writing this from a marketing and design perspective
as well as from a technological perspective--it's my prerogative.  I want our
management tools to kick the hell out of the competition, both for the glory of
it and (more importantly) for the fun of it.

So, please don't mistake my intentions--I'm not telling you to "fork off and
write your own".  I'm holding the door open to see what you can do, and I
promise if your infrastructure can support the two most important features I
noted above (and, as a designer, I'll warn you there are well-known aesthetic
limits to the JSTL) I will be happy to join efforts with you down the road.  For
now, I suggest you code like mad, keep your efforts in a CVS repository
*somewhere* public, and keep us all informed on your progress.  I'm looking
forward to seeing what you come up with.

N. Alex Rupp (

P.S.  MN is stacked with Geronimo developers.  Midwest Represent ;-)

----- Original Message ----- 
From: "Ryan Sonnek" <>
To: <>
Sent: Friday, October 31, 2003 11:44 PM
Subject: Re: console-web

ok, here are my ideas.  please let me know if they go against any design you
had in mind.

1.  remove html from JSP custom tags.
2.  use JSTL in jsp tags to output lists of JMX services
4.  use xdoclet to generate web.xml and taglib.tld.
3.  refactor majority of code into reusable JMX client for use with other
"consoles".  i think this might work best to add a new module, so that the
console-web module has a dependency on the new module.  this would allow for
the taglib.tld to be picked up automatically without having to specify it in
the web.xml

i think that was pretty much it.  i have not done any development for
geronimo, but have gone through the wiki extensively.  what would be the
best way to go about this?  do the development, create a patch, and submit
it to JIRA?

if i have questions, should i email you directly, or would the dev list or
wiki be a better communication spot?  Oh, and i wanted to let you know that
it's great to see other minnesota based J2EE developers!  i'm from mankato
myself, and down here, J2EE is still in it's infancy.

Ryan Sonnek

"God is the Lord of angels, of men - and of elves."
- J.R.R. Tolkien

View raw message