Saturday, November 8, 2008

Eclipse kills open-source SOA projects

The Register has an article about some SOA projects being killed in Eclipse.

This is exactly why IMO the Eclipse Foundation model of open source is broken. Basically, you can buy yourself a board seat in the Eclipse Foundation by putting up cash - on the order of $250k/year I believe .. so not for us little guys for sure. But then, does money guarantee that a project will attract developers and users?? Of course not!

I'm of course biased, but I think the Apache Software Foundation principle of "community over code" is critical to making successful open source projects.

6 comments:

Lahiru Sandakith said...

Totally agreed.. Dr Sanjiva.. Couldn't said better ..

Poor STP proposal went down plus transfer some to WTP .. Interesting ..

+1 on community over code.

Chris Aniszczyk (zx) said...

The projects weren't "killed," they were simply archived. If there's no community being built around the project for a couple years, there's no reason to have it stick around.

There any many projects at Eclipse that aren't represented by people who have board seats. There are project represented by universities and individuals. For example, the Eclipse Communication Framework (ECF) project is used within the Eclipse SDK and is mainly worked on by a couple individuals.

Apache has a similar process for incubating projects too. They get "terminated" if they don't produce over time.

Also if you look at the Eclipse commits for this year so far, a good chunk comes from individuals.

Ian Skerrett said...

It would seem you might not really understand the Eclipse model. Any body can start an Eclipse project, you don't have to be a member and certainly not on the board.

I actually think killing projects is a healthy sign of a well managed community. Not every new idea is a good idea, so it is important to end of life projects.

btw, I think the Apache model is a great model too.

oisin said...

Hi Sanjiva
Looks like it was a slow day at Vulture Central. I'll reuse the title of my post there - Eclipse doesn't kill Projects; Projects kill Projects. There's no drive-bys from the Eclipse Foundation.

In the STP case, those two components (they are not Eclipse projects, officially) I mentioned in my blog entry [0] just ground to a halt - the community evaporated slowly over a period of years and we were left with code and no active committers. So, no project - community is the lifeblood and if that goes, well there's no point in misleading people that there is life there, so we archive it.

In the STP case, the PMC voted to send those components to the archive (thread at [1]) rather than have people think that they are still operational.

You're familiar with the Hallowe'en event that happens in some countries - you'll see that my blog entry is themed in that vein. Although a bit late in terms of timing :)

[0] http://oisinh.wordpress.com/2008/11/06/open-source-not-proof-against-eol/
[1] http://dev.eclipse.org/mhonarc/lists/stp-pmc/msg00350.html

Sanjiva Weerawarana said...

Whow, my apologies for annoying the Eclipse people! That was not the intent at all.

The title of my blog was the title of register article - don't shoot the messenger!!

You're right that Apache has a process for how projects are killed (or sent to the attic) too.

However, the big difference is that Apache will not allow a project to be started at all with just a few people from one company. (Or even a lot of people from one comapany.)

For example, there's a conversation right now on the incubator-general mailing list about graduating the CouchDB project - and you'll see that some of the concerns are that the PMC of 5 is too small.

Apache's rules are designed to make projects that have a good chance to succeed. Of course not all of them will.

I don't want to further infuriate the Eclipse people, but I don't believe that community health is the critical criteria that Eclipse uses to start projects: It is a fact that buying a board seat gives you the right to start a top level project, right?

Ian Skerrett said...

It is not the title of the blog but this statement 'This is exactly why IMO the Eclipse Foundation model of open source is broken.' which is a bit annoying. You might want to better understand the Eclipse model before saying it is broken.

It is fine that you prefer the Apache model; lots of people do. Just like lots of people like the Eclipse model. IMO both communities are very successful and thriving.

btw, no, a board seat does not allow you to start a top-level project.