Re: [RANT] Maven is both heaven and hell

Wendell Beckwith wrote:

 > You're like
> original band members, but it hurts to say that you all are getting your
> asses handed to you by orgs like Spring and Eclipse.  There just doing a
> far
> better job on the dcomentation and website.

Having used maven1 for a long time (and having been blown away by the
concept of a build system that "already knew how to do stuff in a
mutually agreed way", replacing "yet another half written custom ant
script"), I decided that it was time to sell the current project team on
the idea of maven2.

The conclusion of the attempt to use maven2 is that it is simply not
finished yet. Some features taken for granted in maven1 are
missing/incomplete, and the documentation is missing/incomplete.

I think the maven2 project is showing signs of the second system effect
- maven1 was carefully and thoughtfully constructed, documentation
carefully and thoughtfully created. And - it helped that maven1 was
largely complete before people discovered the concept of an intelligent
build system.

maven2 seems to have been built with enthusiasm - but crucial elements
(like key maven1 features, and documentation) have not been completed.

Luckily, there is no evidence of the second system effect in the design
of maven2 (IMHO of course), the problems are in the finish of the
software, meaning that fixing this means altering the focus from new
features to finishing existing ones, and completing the documentation
(as opposed to revisiting a design, or rewriting code).

The reason this is important is this:

maven1 was a complete no brainer to sell to projects. Once I had shown
people that there was no need to construct ant scripts to do everyday
tasks, maven1 "just knew" how to do things, and this was a huge win,
case closed.

On the particular project I am on now, maven1 was considered and
rejected for not supporting transitive dependencies (fair enough) so
they cooked up their own half working ant scripts, using ivy to handle
dependencies. maven2 does support transitive dependencies, so in theory
it should have been a no brainer sell, as before. But in reality my
testing the waters has uncovered a miriad of problems, leading us to
suggest that maven2 initially just be used to generate documentation
(mvn site).

I agree with comments that the documentation needs urgent work, and I as
a new user of maven2, have been trying to add what I consider missing
information from a new user point of view to JIRA (ie, what information
would have helped me use maven2, that was missing or incomplete).

If users could channel issues causing them frustration with the docs
into concise JIRA reports "I am trying to perform task X but the docs
don't tell me how" (which needs to be done at the time, because after
you finally figured out the problem, suddenly that JIRA report doesn't
seem so urgent any more), it will go a long way to indicate to
developers where there are gaps that need filling.

Regards,
Graham
--