Le jeudi 5 janvier 2017, 22:40:10 CET Robert Scholte a écrit :
> Hervé Boutemy:
> It seems that all issues have to do with expanding URLs and connections.
yes, calculating effective values for each pom

> Right now there's business logic in the ModelBuilder to calculate such
> values. We already had some discussions about this :) My question is: is
> this really necessary for the effective pom or are the related plugins
> (m-site-p, m-release-p, etc) capable to calculate it?
a lot of things can be done independently by plugins: that's a fact. When you 
look at it, report plugins configuration and execution is now in maven-report-
exec, which does a lot of things similar to what Maven core does to run Mojos.
It makes things really complex for plugins, to redo something that was until 
now done centrally by Maven core.
Then honestly, is it necessary? it's a question of choice.

> For the SCM
> connections it is quite clear for that the SCM provider should
> (re)-calculate the connection, and don't use the SVN concept as done right
> now (in fact, you can find scm-specific hacks in the code)
no, it's not that clear: yes, we can make something very open (more open than 
url(child) = url(parent) + / + artifactId), then complex, then require to 
delegate, then don't have the effective value in effective pom and force 
plugins to calculate effective value when they need it.
or we can consider that the little improvement I did on the algorithm (be able 
to use a path instead of artifactId, or not add anything) is sufficient, since 
adding a path is really something you expect from url semantics

I'm convinced of the second solution, and we both know didn't convince you yet 
:)

> I really hope it is not necessary to introduce these special properties.
they are the minimum to have something a little bit configurable but yet 
standardized

> I
> have more hope in the distribution-pom or leave the responsibility to the
> plugins using these values. For that reason no 3.5.0 for me.
I don't like to postpone such work done for one year, but since 3.5.0 is on 
convergence and mutual understanding, it's clear I'll need to accept to 
postpone...
I'll prepare a branch and we'll have a discussion, probably after 3.5.0

Regards,

Hervé

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to