incubator-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Robert Burrell Donkin" <>
Subject Re: How strict should podling release reviews be?
Date Wed, 03 Oct 2007 20:02:08 GMT
On 10/3/07, ant elder <> wrote:
> On 10/2/07, Robert Burrell Donkin <> wrote:
> >
> > On 10/2/07, Noel J. Bergman <> wrote:
> > > > i'm thinking more of a benevolent educator than traffic warden
> > > > style reviewer role.
> > >
> > > When it comes to legal issues related to a release, the warden role is
> > the
> > > more appropriate.  It benefits neither the project nor the ASF if we are
> > lax
> > > in that regard.
> > >
> > > Some of the things that they need to do are identified by RAT, and would
> > be
> > > non-issues if they would correct their build process to do them
> > > automatically, e.g., inserting the license and disclaimer files where
> > they
> > > are supposed to go.
> >
> > i do believe that there's a definite problem here. there's too much
> > energy wasted by everyone.
> >
> > the IPMC cannot actively oversee the code bases without automation.
> > so, the only real oversight happens at release time. this is bad for
> > everyone. really, we need to automatically scan and analyse the
> > incubator codebases.
> >
> > i hope that may help
> That RAT proposal looks really good, its just what we need. I can't promise
> to contribute much code but i'd definitely hang around and help test it on
> things.

hopefully it will be easy to contribute in small ways without too much
effort. this is particularly important since a lot of meta-data needs
to be collected. this probably isn't feasible without active help from

for example, a good guessing algorithm for generated files needs good
meta-data about the ways common programs mark files as generated. so
release managers can contribute by submitting new patterns whenever
RAT doesn't correctly recognize a generated file.

another example, discordia aims to collect meta-data allowing
artifacts to be matched to license meta-data. when release managers
encounter a jar (or other binary artifact unknown to discordia) they
can submit meta-data.

> Until that gets implemented (or maybe as part of its design?) could there be
> a wiki page documenting each rule RAT would check?

RAT just automates tedious checks that reviewers carry out by hand.
again, this is going to require collection of meta-data analysis rules
for automation.

> That way  we could have a
> complete list of each specific requirement in one place to make it easier
> for both podlings and reviewers to check manually till RAT is done. If we
> had such a list then it could be only the things documented there are
> release blockers, or at least if a release is blocked the reason should get
> added to the list so we eventually have a fairly compressive list of the
> rules so everyone knows what to expect.

different people have different ideas about what are blockers and IMHO
this is good

i've seen very few -1's, what's much more common is for people with
criticisms to post them and not offer a vote

i would expect a -1 only if the apache policies were broken

> I'd have a go at creating such wiki page with the rules I know about if
> people think its useful but i expect others would need to help out if its
> going to get very comprehensive :)

IMHO the wiki is just a distraction: the real problem is that the
release management page is very unfinished. if there are people with
time then improving would be great. volunteers?

- robert

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message