Not that I should have much of a role in this discussion but I'd like to contribute some thoughts stemming from an offline discussion I had.
I think this discussion is still missing the point. There are a lot of outsider articles on "what is wrong with Apache" these days, most of them refer to the total disinterest (by many developers in the projects) on "the market" meaning what do the user's actually want. I'd say this is a component. (Please take this as somewhat of an outsider who has a lot of experience with Apache work-products) (as a symptom of this: Apache is OBVIOUSLY a better Web Server, TOMCAT is obviously a better App server of sorts, and though not a apache project JBOSS is a great enterprise server....so why is IIS gaining ground despite its overall suckiness?) The second component is an overall lack of unity-of-purpose. XML.apache.org hasn't reached a critical mass and in my opinion may never because it does have unity-of-purpose and I think that is part of why Stefano recommended I approach Jakarta first. POI has a lot to contribute to XML.apache.org but it has a lot of stuff that *would* contribute more to Jakarta's purpose if it had one. This isn't a slam, hear me out. The Apache group had a unity of purpose early on. They had a product: a webserver. Everything that Apache did had something directly to do with that product. Some things were semi-independent so subgroups seemed like the best way to handle it. Java-Apache had this unity-of-purpose: Java on Apache. Well for Java on Apache you need a mod to handle that (since everything is a mod in Apache) so you get mod-jserv, of course you have a lot of things that roll in and out of that based on serverside components for developing with your java mod. But you have unity-of-purpose. (or at least for a time) What is Jakarta's mission? "server side java" stuff. What is your product (look at the homepage)....whoa thats a big list of subprojects... Wait is ant a server side java tool? Well..kinda sorta (build server)... what kind of server-side java stuff. XML.apache.org has a few well-defined "products" with the "main" one being Cocoon. This may change slightly as the web services thing comes to a head (as the speaker coordinator for my JUG www.trijug.org I can tell you this is coming to a head) and more web-servicey things happen with XML rather than publishing (Cocoon) and maybe at that time there should be a webservices.apache.org (and webservices will expand beyond XML), but for the moment you've got real products and a unity-of-purpose. (Which parts of POI fit well into..the cocoon 2 serializers for instance and others do not) So what do most people (users) come to Jakarta for? Tomcat. Why? Go to the front page. A big rattled on list of components....If I don't know what I'm looking for suffice to say I won't find it. If I say "Tomcat" the general IT population knows what I'm talking about. (and the rest know what I mean if I say "the successor to JServ) Here's my 2c worth (and unless asked its the only thing I'll contribute to this discussion): Defined unity of purpose: sever side java is now too fuzzy of a mission...what are your products and categorize them: application server (tomcat) build and development tools (ant/log4j/etc) document management and publishing (lucene, POI, etc) application frameworks (avalon, struts) The Apache brand is worth a lot. You say Linux in a corporate environment you get a dirty look (once upon a time we just said "Solaris-clone" and installed linux to avoid political battles ;-) ), you say Apache you get a less dirty look. (You're still a radical but IBM and Sun said you're an okay radical). Jakarta needs to do some actual PROJECT work. Go in and pull these disparate components into distributions (Redhat doesn't point you to their website to download X, and then GLIB and go try and put it together yourself...not that Jakarta should be redhat, but the point being having distributions). This helps create unity of purpose as things start going into distributions and distributions generate "requirements" and "needs" which roll into features. > I think this equation misses the important second order affects of > collaboration. >My feeling is that communities need a critical mass, as do >meta-communities. I'm not sure what that size is, but in my mind the >XML >project has not reached it. And it is not just size that is the issue, >it >is the presence (or lack) of people who cross pollinate ideas. Within >Jakarta, this is being done on an ever increasing scale. I think you need to have points. Both to discussions and to work. In the POI project. Try this: submit to the poi-devel list (and Marc, Ken and "the lurkers" may not be monitoring so the experiment would be fair) a proposal or patch to do something obviously outside of POI's mission (crack those MS file formats right open, provide apis and XML publishing utils for those formats).... I bet you you'll get a unified "thanks, hey why don't you start this somewhere else on sourceforge for instance...we'd love to help you integrate POI into your project, but I think this falls a little outside our mission." I could submit just about anything that was "Server-side" and Java...that's vague. >Now for a thought experiment: if POI were added to Jakarta, would this >metric overall increase or decrease for Jakarta? If POI were added to >XML, >would the metric overall increase or decrease? I would say POI isn't really relevant to this part of the discussion. POI might catalyze some cross pollination that wasn't there. I would say POI would help increase the user-base for a number of projects including Cocoon, Lucene and Tomcat. I'd say a third component may be that there are people who use ad hominim attacks for which there is no justification (Please read any textbook on the principles of Argument), there are a number of circular arguments that have no possible conclusion and side-shows arguing "coding styles" is an age old way to increase the acrimony in any software development group. If you're on the PMC especially you should provide leadership and aim the discussion toward things that further the project as a whole and never engage in arguments that have no logical conclusion. "Should my brackets go on the same line or the next" is a good way to just tick people off. Discuss the finer points...will the argument ever have a logical conclusion. What about coding standards. Who in their right mind would enforce them? Would enforcing them truly serve any purpose? Would enforcing them be worth it (and how much acrimony would result?) Lastly, document. When you're done documenting....document some more. One of the problems is that without documentation, Apache developers don't know what you have...let alone your user community (and they aren't going to even look as hard!).. I know I know...opensource and documentation. If there were one coding standard I'd enforce is I'd not accept new features/subprojects that didn't come with documentation. Just Remember, the view only looks so grim to subscribers of this list. Within the subprojects there is much less chicken-little-ism (the sky is not falling) Anyhow, this has nothing to do with a comment on whether POI should be a Jakarta subproject. Why would I want it to be? Because I think it might be the best way to further my own objective. (Force IIS off of every server, Force tomcat on, with maybe some JBoss and Linux) I think it also would further both some Jakarta projects I'm interested in and Cocoon. Having it at Jakarta would raise awareness that one can have their Unix, Java and yet still make your accountants happy. I realize these are user-minded needs..but users drive requirements which should drive features. You pick your user community with your unity-of-purpose and mission. (There will be no C# components added to POI... unity-of-purpose...mission, though some users might like it if there were). POI fits well into Jakarta's stated mission of promoting server-side Java (maybe thats a better measuring stick...its not only serverside and Java....it promotes serverside Java). Would it divide the community? Wow a library to read/write the most popular document formats used in business divides the community. That would make me wonder if you had a community in the first place. > >Ignoring the fact that the following are clearly related, which is more >important: community coherence or mission coherence? IMHO, the latter drives the former. Those who do not initially conform to the mission will disappear or assimilate. Anyhow, I'm just a simple Java programmer and multi-apache project user. These are my thoughts which I thought might help. If not, my apologies, if so then I'm glad. -Andy -- www.sourceforge.net/projects/poi - port of Excel format to java http://developer.java.sun.com/developer/bugParade/bugs/4487555.html - fix java generics! The avalanche has already started. It is too late for the pebbles to vote. -Ambassador Kosh -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
