On 2015-07-27 01:20, Roman Shaposhnik wrote:
I'd like to raise a somewhat orthogonal point. Mainly the fact that our
obsession with doing good work with podlings could, very well, be
obscuring a much more important issues. And given how limited
our resources of eyeballs looking at releases are -- we need to
be practical.
Now, while I couldn't agree more that IP hygiene of releases is one of
the corner stones of what makes ASF what it is, I have to be realistic in
accepting the fact that once podling is out of the incubation and becomes a TLP
the level of external scrutiny drops by 90% and all bets are off. Some TLPs
do great releases. Some do really poor ones. Sometimes ppl notify the board.
Once again, I can't be happier that there are folks like Justin who spend
an enormous amount of their personal time helping to vet releases of
podlings. I only ask: should his (and guys like him) time be better applied
to helping foundation make sure that our TLPs are doing what they are
supposed to do as well?
I too am very delighted that we have volunteers who are both willing and
able to help us through these somewhat dry and dusty areas of releasing
software. I believe our time is better spent if we educate new podlings
on proper release and IP procedures, so as to prevent these things from
happening later on when they are released into the apache wilds with the
273 other projects, and as such, I think it's better to have these
checks (and the people doing the checks) in the incubator. Having said
that, I do realize this puts a lot of pressure on very few people, and
the asynchronicity of going back and forth between podling and incubator
is not the optimal way. But I believe this is something the incubator
needs to teach podlings.
What if, for example, we were to come up with a more automated and
educational system for this, so as to get it faster through the human
control? We already have Rat that does most of this, so couldn't we make
a "super rat" that does a bit more, specifically designed to educate the
incubating podlings, maybe a web interface where you just throw a
tarball at it and it'll tell you what needs to be improved.
Sure, it would have a significant cost in people hours at first, but if
the Incubator is going to be around for another decade or more, it would
definitely pay off in the end and make initial releases faster.
I'd be happy to collaborate with others on this, if there is an interest
in giving it a go.
With regards,
Daniel.
---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscr...@incubator.apache.org
For additional commands, e-mail: general-h...@incubator.apache.org
