Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
[m2e-users] Long term outlook for 'M2E plugin execution not covered'

Hi all,

I recently upgraded to the new m2e release that has arrived with
Eclipse Indigo and I'd like to give some feedback.

Like many users attempting to use m2e, I've had problems with the
behaviour described here:
http://wiki.eclipse.org/M2E_plugin_execution_not_covered

Firstly, I found it a bit disappointing to arrive in Eclipse Indigo
with my first Maven project imported, only to be confronted by a
broken build that had to be fixed by heading to Google and tracking
down a somewhat confusing workaround. For devs working in
organisations that build many, non-trivial Maven modules, hitting this
problem early on is inevitable and the experience is not great.

Now that I think I grok the above linked wiki page, I'm very
uncomfortable with the implications that it has for m2e users and
Maven plugin developers. We're now in a situation where:

1. Maven plugin authors need to worry about which IDE their users use,
they need to provide explicit Eclipse integration along with their
plugin. It feels wrong that when writing a Maven plugin I need to
worry about IDEs (rather than simply Maven itself), an indication that
m2e has taken the wrong path here. Where plugins are concerned, m2e
appears to effectively no longer integrate with Maven, it needs an
additional 'integration layer' to be written across the plugin
universe. I'm confused and amazed by this design decision.

2. Projects using Maven need to add IDE-specific (Eclipse-specific)
configuration into their pom, or worse individual developers that work
on a project need to customize their local sources before Eclipse can
even build a Maven project. Bug 350414 should go some way to address
this if fixed right, but no doubt Eclipse will still allow the
lifecycle mappings to the pom where they really do not belong. In the
short term (where the POM is the only place these mappings are
supported), I believe Indigo/m2e is not viable for many, many
projects, since adding IDE specific configuration will be deemed
unacceptable by project owners.

These problems are really bad news for organisations that maintain
many hundreds of Maven modules. It seems like we're in a big mess at
the moment and one that will force users to stick with Helios &
m2eclipse 0.12. I fear that a fundamentally harmful design choice has
made m2e (and therefore Eclipse) less tenable for Maven users. The
case for migration to Netbeans or Intellij has just been given some
solid ammunition.

I'm afraid I don't have an alternative solution to discuss with
m2e-dev since I'm not familiar with the internals of m2e. As an end
user though, I feel there are some big problems here.

Is there any chance that for Indigo SR1 or SR2 the m2e dev team will
revisit this lifecycle integration in a fundamental way that will
address these problems? Maybe it's possible to learn from the Maven
integration done by other IDEs?

I hope this doesn't sound like an attack. No doubt the latest version
of m2e brings a whole load of benefits that I haven't yet had a chance
to experience, and for those I'm really grateful to the development
team for the effort put in. I'd love to get the benefit of those
improvements, but problems I've described here are show-stoppers for
me and probably many others. Does anyone else feel similarly?

Cheers

Joe


Back to the top