incubator-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Craig McClanahan" <>
Subject Re: Killing the incubator m2 repository
Date Sat, 17 Mar 2007 04:38:19 GMT
On 3/16/07, Bruce Snyder <> wrote:
> On 3/16/07, Davanum Srinivas <> wrote:
> > Sorry, i lost you. this whole "we need podling artifacts in central
> > repo" because right now you are putting our user through a meat
> > grinder has no basis in fact. Am asking for JIRA issues, email threads
> > that show that this is indeed a serious issue and not just a made up
> > issue. Show me the evidence is what i am asking.
> By comparison, I've now asked three times if there was any specific
> situation that necessitated the policy of a separate repo for
> Incubator artifacts. Thus far, I've received no answer whatsoever. Was
> this policy simply plucked out of thin air (i.e., a made up issue) or
> was there a specific situation that prompted it?

Part of the original reasoning was that podlings could not make an
"official" Apache release because they weren't actually projects yet,
and we didn't want downstream users to assume that they were.
However, it's pretty tough for a podling to build a community if they
can't release anything at all -- so a compromise mechanism was set up
that required podlings to use things like "-incubating" in both the
version numbers and filename artifacts of their "releases", *and* to
avoid using standard Apache release mechanisms ... essentially, a
podling release is more akin to a nightly build that you might grab
from p.a.o/builds, meaning that you deliberately have to look there
for it.  For Mavenites, the corresponding concept was a separate
repository that downstream users would (theoretically -- as has been
pointed out, it doesn't really work for transitive dependencies) have
to explicitly indicate their willingness to depend on incubator
podling releases, by adding the incubator repository to their POMs

But times change.  Apparently now people think that since the
Incubator PMC has binding votes on podling releases, then these things
*are* indeed official releases, providing both the usual legal
protections to PMC members who voted for it, and assurance to
downstream users that things like code provenance have already been
vetted.  If we buy into this philosophy, then a restriction on
publishing artifacts to p.a.o/dist and the mirrors, and having a
separate incubator repository for them, are ridiculous and should be

My grumpiness is over the fact that *only* removing these two
restrictions only goes half way.  If podlings can create official ASF
releases, then they should also not be subject to any restrictions
like "-incubating" in artifact names or version numbers.  They are
either official releases or not.  They can be relied upon by
downstream users or they cannot.  If we're going to go this direction,
go *all* the way, and erase any distinction between a podling release
and a TLP release, since we would be claiming that there is no legal
difference between them, nor any potential concern for downstream
users about code provenance.  Anything in between (such as only doing
what the vote proposes) leaves us sending totally confusing mixed


> Bruce
> --
> perl -e 'print unpack("u30","D0G)U8V4\@4VYY9&5R\"F)R=6-E+G-N>61E<D\!G;6%I;\"YC;VT*"
> );'
> Apache Geronimo -
> Apache ActiveMQ -
> Apache ServiceMix -
> Castor -
> ---------------------------------------------------------------------
> To unsubscribe, e-mail:
> For additional commands, e-mail:

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

View raw message