maven-m2-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Maczka Michal <>
Subject RE: cvs commit: maven-components/maven-core/src/test/java/org/apa che/maven/plugin
Date Wed, 16 Feb 2005 13:11:26 GMT
> > Other remarks:
> >
> > a) If I understand how lifecycle execution works - all 
> phases which are
> > proceeding given phase are executed before if that phase 
> was requested.
> >   In that case does it mean that to deploy things to remote 
> repository
> they
> > have to be first installed in local one?
> >   It probably makes sense but some issues related to 
> SNAPSHOT artifacts
> must
> > be then taken into account so both local and remote
> >   artifact share the same version.
> All deploy goals must depends on install goal. It's necessary to run
> correctly a build with reactor if the remote repo is down.

I am not sure if I understood you - but my only remark was: this is not how
it works in m1 :)
I am almost sure that this has no negative consequences :) 

> >
> >
> > b) How secondary artifacts will be supported by this lifecycle (e.g.
> > ejb-client)? Will the "package" phase be allowed to produce 
> more then one
> > artifact?
> For ejb, I think that is the role of ejb plugin. package 
> phase doesn't know
> the internal process of a plugin.
I still don't have a strong opinion how secondary artifacts are going to be
Specifically I just don't know if there always be only a single plugin (e.g.
ejb) which will produce
more then one artfact (e.g. ejb, ejb-client) or if there will be more then
plugin involved in that.
Also I don't know how many secondary artifacts might be actually needed
(will we ever need more then 2 of them in this lifecycle?).
Probably the difference between the main artifact and secondary artifacts
will be that the plugin which produces the first one
is "activated" automatically (project type is used for finding that plugin)
while plugins which are producing secondary artifacts
must be somehow activated "manually" so their "participation" in the process
can be taken into the account.

I suppose that for other lifecycles (e.g. site, distribution) it will look
differently as for example we might have 
two distribution artifacts (zip, tar.gz), site war, site zip or even each
report can produce an artifact which may be published
to the repository (e.g. javadocs). I would keep those other lifecycles out
of discussion at the moment as I feel that this can
be generalized later on.


View raw message