Hervé BOUTEMY wrote:
the rationale behind not going directly to 3.0 was that site plugin is hard to
test, particularly now that it is both compatible with Maven 2 and Maven 3,
which is something really new and probably tested by only a few of us
I don't quite agree with this rationale. Ease of testing is not a
criterion for version naming IMO. The main criterion is how many *known*
bugs and missing features there are left. So what are the open issues
that we are aware about? If there are none or only a few, then let's
call it final. If the people who are working on the release feel that
the stuff is stable (which I do) then why not release it as such?
sure, 3.0-beta-4 should at least be 3.0-RC-1, but perhaps not 3.0 immediately:
I'm pretty sure we'll find some important problems when a lot of people try it
seriously
The most efficient way to get people to test something, is to release it! :)
There are real important factors to test, which makes a lot of combinations:
- Maven version: 2.2.x, 3.x
- OS
- phases: site, site-deploy, site:stage-deploy (run? jar?)
should all be covered by our ITs:
https://builds.apache.org/job/maven-site-plugin-2.x/
https://builds.apache.org/job/maven-site-plugin-3.x/
https://builds.apache.org/job/maven-site-plugin-3.x-m2/
I am aware that there are some important differences though, (some ITs
are skipped with m3, or executed with different parameters), which would
be important to review and document I guess.
- deploy protocol: scp, webdav
not really a site-plugin concern, rather wagon
- report plugins used: I don't know how to describe without being a mess...
We (devs) cannot test everything, even the more important it is to get
user feedback.
But at least, with maven-site-plugin 2.3 being out and almost equivalent
(particularly when it comes to Doxia and Doxia Site Tools), we have a clear
line to check if a problem with 3.0 is a regression from 2.3 or not
so this would rather be an argument in favour of 3.0...?
Then I'd better be for 3.0-RC-1 for the moment.
I will support whatever the release manager decides, but I would prefer
3.0-final with a number of bug fix releases following, rather than an
open interval of [RC-1, RC-2,...). More people will test the final
release and there will be more pressure on us to push for bug-fix
versions (which is good! :) ).
-Lukas
Such a discussion happened a lot of time in the past: 3.0 and 3.0-RC-1 are
good choices, but not 3.0-beta-4
The release manager can choose and I'll be with him.
But IMHO we need to ask for people to tell what conditions they tested.
Regards,
Hervé
Le mercredi 6 juillet 2011, Olivier Lamy a écrit :
No objections from me.
beta cycle has started long time ago.
2011/7/6 Lukas Theussl<[email protected]>:
Any objections to making this 3.0-final? AFAICT the plugin is
functionally (almost) equivalent to the 2.x trunk version (only
exception is MSITE-484?), so why keep the beta?
-Lukas
Dennis Lundberg wrote:
Hi
What's the status on this? I know Hervé worked on extracting a shared
component (maven-reporting-exec) for the Maven 3 specific parts of the
plugin. Did you finish with that?
I would like to push for a release of Site Plugin 3 shortly. The only
issue left according to JIRA is this one:
http://jira.codehaus.org/browse/MSITE-560
There are a lot stuff fixed already, and we need to get this out so that
Maven 3 users can benefit from them. Do we want/need to add anything
more before the release?
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
