On 06.04.2009 10:40:51 Carsten Ziegeler wrote:
> Jeremias Maerki wrote:
> > I don't want to be too negative, but I've thought a little about
> > Sanselan. What happens if someone comes up with a 100% Java JPEG or
> > JBIG2 implementation? Or ImageIO codecs wrapping the implementations
> > here? Sanselan is a collection of related items. When one item is
> > finished there's not much incentive to stick around and do further
> > development. Once you got the feature you were looking for, you're gone.
> > 
> > All we have is a few wishes, but not enough time to make certain
> > things happen.
> > 
> > Philipp is basically hinting at part of the problem: "ok, jpeg is
> > missing but can be worked around easily." ImageIO offers so much that
> > you only need Sanselan in very special cases. JPEG support is not as
> > badly needed as it is available on every Java VM >=1.4. You (maybe) extract
> > the metadata with Sanselan but load the image with ImageIO.
> > 
> > ATM, I have trouble seeing how a community can be built like this. Of
> > course, a small group can allocate some time to work on a few things and
> > get the project through graduation, but isn't it then only destined for
> > the Attic afterwards? Please don't get me wrong: Sanselan is a great
> > library and offers some very nice functionality. At the moment, I just
> > don't see it graduating any time soon. The best I currently see for
> > Sanselan is to become an Apache Commons subproject that is allowed to go
> > dormant for some time and to be reawakened from time to time to add some
> > functionality. However, I fear that Sanselan would get lost in the noise
> > of the joint mailing lists. But Sanselan as a TLP? I have trouble
> > imagining that. For the third time in a row, we basically report almost
> > the exact same text.
> > 
> > I'm curious about your ideas. I'd like Sanselan to succeed.
> > 
> Ok, so here we go:
> I think basically we have three options (I'm just listing all of them,
> I'm not a fan of all of them!):
> 
> 1) continue in the same manner and hope that at some point of time the
> community will grow and we can graduate - but I fear this won't happen.
> Sanselan is a cool and great library which is perfect for the things it
> does. It's easy to use it and it works. Which is of course great for a
> software but bad for community building :)

Right, which means this is not leading anywhere.

> 2) close down the project and call it "failed incubation" - now this
> would be a really bad and in the end it wouldn't help anyone. Code like
> this are a great benefit to the foundation.

Yes, it would be a shame for all the work that has gone into making the
incubation happen. But if Sanselan went on outside the ASF, we could
still use it.

> 3) try to graduate :) Now, obviously we can't make it into an own TLP,
> so we need to become a sub project of an existing TLP. I thought about
> Tikka (well, Lucene in this case) and Apache Commons. I mentioned
> Commons to Craig and he thought that this would be a good option (I hope
> I got this right Craig) and you're also coming to this conclusion.
> Apache Commons is imho a perfect fit for software like this; it's a
> focused domain, with working/perfect code and it might just need some
> maintenance from time to time; I guess most of these criterias are true
> for several projects in commons.

I think the Commons option is the best compromise. Burying it in
Tika/Lucene isn't a good idea IMO (visibility).

There's a similar case with an ASF-wide XMP library: JempBox "hidden" away
in PDFBox and another "hidden" in XML Graphics Commons. The best way is
to consolidate in an Apache Commons subproject.

> Maybe there is a fourth option?

I don't see one.

> But I think we should talk with the Commons people and see what they
> think about it. I think we met all graduation criteria apart from the
> diversity/community aspect. So we're ready to graduate into a sub
> project like commons.
> 
> WDYT?

+1 to trying to migrate Sanselan into Apache Commons Sanselan.

> Carsten
> -- 
> Carsten Ziegeler
> [email protected]




Jeremias Maerki

Reply via email to