Re: [Wonder-disc] Wonder "fake releases" frequency (Maven related)

[Mike S. <ms...@md...>]

> I completely agree. :) That is the best solution for the general  
> problem
> related with the Wonder release process. Not only Maven users, but all
> Wonder users will be benefited by this solution.
Ideally, yeah, you would depend on a particular version of  
ERAttachment and that would just have its own cascading deps.  I think  
in the majority of cases all the deps would match version numbers, but  
i can see that maybe there's some particular bug fix to ERAttachment  
that falls at a time when ERX has a bug.  This is, though, pretty rare  
(problematic things TEND to be fixed pretty fast).

> The problem is: Lachlan and me don't have the knowledge nor the
> authority to define when is time to make a new Wonder release.
I don't think you're defining an "authoritative Wonder release", I  
think you're just marking points in time of relative quiet in the  
source tree ... For instance, there haven't been commits to Wonder  
trunk in several days.  It's a fairly safe assumption for anyone  
watching the commit log emails that you could mark a version.  If you  
wanted to make this "smart" you could probably write a script that  
does releases after x days of quiet since the last commit.  You could  
manually override it if there was some particularly busy time, but I  
would venture to say that if there hasn't been a commit in, say, 3 to  
5 days, that things are probably OK.  Obviously this has only slightly  
more of a guarantee than the nightly, but if you're looking for  
something automated, you're probably not going to get a whole lot  
better than that.

> As you
> said, we need "Mike" (read Mike as any Wonder developer capable to say
> that he finished a bunch of useful changes) to tell us that is time  
> for
> a new release. So, this solution demand some changes in the work  
> flow of
> Wonder developers.
It's pretty rare that big new things go without any announcement,  
though, so you can always key off of that + just a little bit of  
watching commit logs.

ms