I can add a bit of detail on the core side in addition to Vincent and
Lukas' mails.
On 03/03/2009, at 6:11 AM, John Casey wrote:
What exactly is the plan for Doxia and Maven 2.1.0? Who's doing what
to push this along?
You should first review the exact same question I asked 5 days ago :)
1. What compelling reason do we have to upgrade Doxia in Maven
2.1.0? I understand that there's a largish assortment of plugins
that cannot upgrade until Maven does...but what is in Doxia 1.1 that
we can't live without until the next release of Maven (by which time
- HOPEFULLY - Doxia would have been released and in the wild for
some time)?
For me, because it breaks the deadlock on the release and facilitates
everything else moving forward rather than the split that occurred.
2. Is there a good reason why we can't publish a minimal set of
interfaces and exceptions that are shared between Maven and Doxia,
such that we could peg Maven to a version of that API and Doxia
could maintain compatibility with that API, and specific versions of
Doxia used in Maven could be a little more fluid?
This is already the case. We really only need the Sink and
MavenReport. The original intent was that those interfaces wouldn't
change, but Sink did. It needs to be decoupled, but for now, updating
just the sink/logging api will facilitate plugins using new versions
(and new doxia releases can be made without changing the sink/logging
that is now bundled)
For example, if the Site plugin were to instantiate the Doxia bits
and pass them through the Maven core to the report plugins...can we
make it so Maven core doesn't have much to do with it, and the site
plugin/report plugins can be the ones to respond to new Doxia
releases?? If we could agree on a minimal API artifact and just get
that released ASAP, it seems like we could upgrade to Doxia 1.1
after the core 2.1.0 release either through plugin-level deps or a
new release of the site plugin.
This is the ideal solution - I tried various iterations but it doesn't
work in a backwards compatible way. Maybe some on-the-fly ASM wizardry
could do it by creating the correct version of the sink interface at
the time, but I think a better bet is to break compat in 3.0 and
centre the sink handling in the site plugin instead as you say.
If we really are stuck with this sort of lock-step release cycle,
isn't there something we can do right now, before releasing maven,
doxia, or what-have-you, that would improve (solve?) the situation
for next time?
I don't think it should be needed again if the API JARs are managed
carefully. The key thing is that the Doxia release needs to start
*right away* as originally suggested.
Cheers,
Brett
--
Brett Porter
br...@apache.org
http://blogs.exist.com/bporter/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org
For additional commands, e-mail: dev-h...@maven.apache.org
