I'll point out that the link Martjin pointed to is a follow up conversation after the vote thread that originally introduced allowing podling artifacts to be distributed via Maven. https://lists.apache.org/thread.html/43c243fcdc50777cb521e39fefed93f14e3ccb5c869bc02447ae68cd@1221028489@%3Cgeneral.incubator.apache.org%3E
Now that I've seen these threads, I understand the perspective our forefathers had when applying these restrictions. And I'll reiterate my concerns with it - we created a policy that applied specifically to maven artifacts, that need not apply to other distribution mechanisms - PyPi, NuGet, etc. Though I'm not sure we've allowed or disallowed such a thing. Anyways back to dissecting the thread. Here are some interesting notes (and commentary): --- https://lists.apache.org/thread.html/a403eb753829ba3cd6a5023ebe0ac866deb9eba1cedefb17a84b4001@1222561027@%3Cgeneral.incubator.apache.org%3E William Rowe, clarifying that an end user shouldn't have to be worried about an incubating dependency. Its on the distributor if it goes *poof*. It also reiterates a point that's been stated - consumers may consider -incubating to be a sign of stability, robustness but in actuality it's purely an ASF term. Here's another note, RE success of podlings. We have 64 active podlings. In total there have been 276 podlings, of which 44 have retired. The concern raised and why -incubating is used is to call out that a podling may fail and retire. There are 177 TLPs. We have 41 projects in the attic, DeviceMap joining shortly (just not listed on https://attic.apache.org/) so that makes 42 projects in the attic. The numbers are about on par for the likelihood of a project failing to graduate as it is to fail as a TLP. https://lists.apache.org/thread.html/de64b98fe7d666f88fb3741bb267e8f523e61d2f9965e1227cb09cb9@1222585509@%3Cgeneral.incubator.apache.org%3E Niclas poses some great points, including voting no to a podling that's clearly struggling. Sounds like a better solution than additional requirements. https://lists.apache.org/thread.html/36f6477b5a211d6a7374fe3b7f5640dc31f9bd3fd0e2be337eecc2ea@1222817549@%3Cgeneral.incubator.apache.org%3E Upayavira makes a claim of something I have long wondered - what exactly do we mean by "endorsed"? "You need to clarify what "endorsed" means. It _is_ endorsed as "opensource", but not as "open development". " This statement is later followed up by notions that the ASF endorses a release when the binding votes pass, and that we make no claims that a podling will stick around. This is even true of a TLP. Though I suspect once a project has reached a certain level of maturity, there's always someone willing to help out. Its even followed up with a notion from Daniel Kulp, where we may piss off users by changing maven coordinates. --- There's a few things that this seems to point out. - We're not approving releases that are way out of line from policies and procedures. We call out when a podling has incorrect licensing claims, is distributing files that shouldn't be distributed. - A podling's release is endorsed, while the project behind that release is not endorsed. - Users of the podling's release I think at this point we should start a vote on removing the -incubating requirement on maven artifacts. I'll wait until Monday at least to start the vote, due to remaining holiday time. John On Sat, Dec 31, 2016 at 11:00 AM Raphael Bircher <rbircherapa...@gmail.com> wrote: > Hi all > > Am .12.2016, 15:40 Uhr, schrieb Daniel Dekany <ddek...@freemail.hu>: > > > The original question in this thread is much less generic than in the > > linked one though, so it's certainly easier to answer. Just as a > > reminder, the specialities are these: > > > > - This project already have suffixless releases from before > > incubation, which are widely used. If the incubation fails, those > > versions won't be any better that the incubating ones. After all, ASF > > won't go around and delete those releases from the Maven Central, > > from SF.net, etc. > > > > - The project doesn't have an org.apache.* Maven groupId, nor a such > > Java package (in 2.x.x, because of backward compatibility, and we > > are talking about 2.x.x only here). So as far as I see it's quite > > unlikely that the user will see "Apache" without "incubating" > > because we have removed it from the *Maven* version number, at least > > if we change <name>Apache FreeMarker</name> in the POM to > > <name>Apache FreeMarker (incubating)</name>. > > > > For my point of view, this would be the right solution. As I understand > (incubating) should be added to the Project name and not to the version > number. > > Regards, Raphael > > > -- > Mein Blog: https://raphaelbircher.wordpress.com > > --------------------------------------------------------------------- > To unsubscribe, e-mail: general-unsubscr...@incubator.apache.org > For additional commands, e-mail: general-h...@incubator.apache.org > > >
