Hei,
I have tried to use the latest selenium plugin and I am facing the
MSELENIUM-22 issue (selenese goal).
This issue exhibits something I feel is dangerous to the maven
ecosystem. To use the plugin I have a groovy layer integrating with
ant calling the selenese ant task.
How much runtime impact does this have on my build (memory, time) ?
Doesn't this become much harder to debug ?
Can't this be simplified ? Shouldn't we aim for simpler integrations ?
When I try to design plugins, I try as often as possible to limit the
amount of layers between me and the target tool. Either I call the
executable, either I call the API, or if I don't have time, I
integrate with an existing ant task.
Ideally I would like to see maven goals sit at the same level as ant
tasks, reusing an build tool agnostic API.
I believe that for maven to be robust and fast we have to push the
"try to have plugins interact as close as possible to their target".
I understand the fact that reusing an existing layer is tempting. But
does it serve us on the long run ?
Thoughts ?
--
Jerome Lacoste
---------------------------------------------------------------------
To unsubscribe from this list please visit:
http://xircles.codehaus.org/manage_email
