Developing with Maven
Subject:   The Waterfall Model Applied
Date:   2003-10-24 02:32:38
From:   ianfairman
Response to: Dealing with the kitchen sink


I'm willing to be persuaded that Maven could be an improvement over raw Ant. But when I look at how Maven is put together it unsettles me. In trying to deal with that 2% I think they've overcomplicated things and forced themselves into making a lot of decisions in a short period of time which I wonder if they'll regret. I've seen this *so* many times and the results are rarely good.

Maybe XSLT isn't the answer - or maybe it's the whole answer or part of it - I don't know. But the "big bang" release of Maven leaves little room to rework design decisions without breaking things - it's the waterfall model all over again.

One final thought, which I came across this morning: Orgel's Rule which states "Evolution is cleverer than you are" ( Start with a worse-is-better solution and see where that takes you - if XP has one lesson to teach us it should be that one at least.


Full Threads Newest First

Showing messages 1 through 2 of 2.

  • The Waterfall Model Applied
    2003-10-24 04:47:03  anonymous2 [View]

    I'm curious as to where, specifically, you see examples of unnecessary complication.

    I'm also curious about why you see Maven 1.0, RC1 a "big bang" release of Maven; as far as I know, they've been the typical open source product with all the early alpha and beta releases, etc. Plenty of iterations occurring, plenty of reworking in the design decisions.

    Could you clarify on both of these points?
    • The Waterfall Model Applied
      2003-10-24 08:11:31  ianfairman [View]

      I suppose I'd see the complication as being in terms of too many features being put in version 1.0. Did Maven really need in version 1.0 a repository (with all its configuration management implications), a scripting environment based on Jelly (why choose that?), integration with Gump, project data visualisation, source code metrics?

      As for the "big bang", most developers are not seriously going to use Maven until it goes final - all iterations prior to that will be used by a more limited community. The developers will get a lot of useful feedback then which would have been easier to work with if Maven had been a smaller system.

      Maybe others see Maven differently. I'm willing to be proved wrong. Only time will tell whether the Maven developers took the right approach.