maven-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Brian E. Fox" <bri...@reply.infinity.nu>
Subject RE: Maven 2.1.0 and Doxia
Date Tue, 03 Mar 2009 14:57:18 GMT
The more this thread goes on, the less optimistic I feel about this going into 2.1. We already
know the 2.1M1 is stable and the point was to get it out in a release that people can use,
ie non-milestone. Making radical changes at the last minute is not good for stability and
not good for the users. I think this should go into 2.2 and there should be a release cut
and integrated into the 2.2 snapshots immediately so we have time to understand the issues.
Using a maven release to effectively test doxia goes against all the progress we've made in
the last year to stabilize the releases and improve quality.

(this is the same argument I think I used in 2.0.9, 2.0.10, and since the release didn't happen,
it's still valid imo)

-----Original Message-----
From: John Casey [mailto:jdcasey@commonjava.org] 
Sent: Tuesday, March 03, 2009 9:51 AM
To: Maven Developers List
Subject: Re: Maven 2.1.0 and Doxia

Lukas Theussl wrote:
> 
> Hi John,
> 
> If you would have used the time it took you to write this email for 
> installing and testing Doxia 1.1 you probably wouldn't have to ask 
> anymore if it's a terrible idea. ;)

In my experience, five minutes isn't enough time to do anything when 
you've got more than one snapshot dependency/plugin you're trying to 
use. It's not trivial to replace a whole stack of artifacts with another 
whole stack of artifacts, then dig up a test project that will put the 
snapshots through their paces properly.

> 
> However, the main point is that I also see no reason for NOT upgrading 
> now (as I do not accept your FUD arguments). 

This is not merely my blowing smoke and trying to inject fear; we've 
been bitten time and again by bad/untested dependencies that get pushed 
out just ahead of a Maven release. I've spent quite a lot of time in the 
past on the front lines after these releases, answering questions about 
why things are messed up and debugging problems for people to find a 
workaround.

I don't think taking a slightly more conservative view of the release 
process as a result is in any way FUD. These are not unfounded views.


A few people have now taken
> their time to test 1.1 and have provided positive feedback (Brett, 
> Herve, Wendy). Doxia 1.1 (then beta-1) was present in the 2.0.x core for 
> 4 months [1], unfortunately it was reverted for the 2.0.10 release [2] 
> for reasons that I do not recall

These reasons were failing tests. I guess we didn't have the test 
coverage in place to figure that out for four months...good thing we're 
all set now, though.

> 
> As for the improvements you propose, it certainly sounds good and all, 
> and IIUC along the lines of what Jason is planning for Maven 3 [3], but 
> as long as there is nothing concrete on the table I don't see why we 
> shouldn't use what we have now (ie Doxia 1.1).

IMO, the only thing we know we have now is the released and 
battle-proven versions that went out with the previous release(s). These 
may have warts, but they're known quantities. We don't have a Doxia 1.1, 
that's kind of the point I was trying to make. We *may* have one soon, 
and maybe it'll work or maybe it won't. Maybe it'll introduce a crop of 
bugs that keep us from saying that this Maven release is *definitely* 
progress, *definitely* better.

Our problems with upgrading Doxia in Maven core are not simply 
prejudice, they have a long history of causing really difficult 
problems. This means Doxia has to be subject to extraordinary scrutiny 
before we include a new release in Maven itself.

-john


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
For additional commands, e-mail: dev-help@maven.apache.org

Mime
View raw message