Swaroop C H

blog books about contact subscribe

How to kill an open source project

27 Mar 2005

... by hiring the top guy into your company and then cut off all incentives for the community to get interested.

Yes, I'm talking about IronPython and in this case, the company happens to be MS.

Read what Edd Dumbill has to say about it, especially this part:

Third-party patches won't be considered until after the 1.0 release. Hugunin encourages people to be involved, but only in filing good bug reports and feature requests. And when doing this involves a .NET passport, and using the GotDotNet web forums rather than good old mailing lists, it's a bit of a disincentive. Added to that, there's some uncertainty about the freeness of IronPython's license. While it looks free, it's got the same name, "Shared Source", as several Microsoft licenses that definitely are not free.

I have nothing more to say except that such an awesome 'open source' project has been effectively killed off - IronPython may go on to become 1.0 but it is now reduced to a one-man show when so many people could have contributed to it.

Comments

Sriram says:

a) See why Jim doesn't want patches *now* (from his mail to the list). He says the code is under too much flux and that he can implement bug fixes more reliably. He's just asking people to wait a bit for code to stabilize

b) Maybe Edd Dumbill should try signing up for a .NET Passport and see that all it takes is a email address and your name,country,etc. If you don't like giving that out, use fake email addresses, names,etc. In fact, Google's login is very close to Passport- but no one complains about that.What would he rather prefer? A different login for every Microsoft site out there? If that were the case, I would have need close to 100 logins right now.

c)I find it funny Edd questions this license based on the fact that it is called 'Shared' rather than bothering to read the contents of the license himself. Or maybe he has - he says "While it looks free".

On the other hand, if people are looking for a FSF-definition of free they may not find it.

Read through the code (and a lot of people have - and seeing the lack of posts about this on the IronPython list, I suspect people haven't been able to find any holes).

Does some code have to be OSI-certified to be "open"? If that is the case, then that's a pretty narrow view of "open", IMHO. Matusow(the guy in charge of Shared Source) talks about this at http://blogs.msdn.com/jasonmatusow/archive/2005/03/10/394009.aspx

Sriram says:

Hey dude - could you bring back your 'subscribe to comments'? Would make it easier for me to keep track of these threads.Also, do make this text box bigger - helps when you're typing in long rants :)

Back to the flames then

a.Legal reasons are a reason *also*. But quoting from Jim


"Given the early and rapidly changing state of IronPython, as a developer
I'd much rather have a clear test case than a patch file for almost any
bug because I can usually develop the fix myself faster than I can
understand and validate an externally submitted patch


Plus - given that people can fork this at any point of time,is this an issue? If someone doesn't like the fact that Jim does the coding, he/she can open up their own fork.

b) I hate the forum idea too - and Jim says he'll leave the list open too for people who prefer the list.Forgetting the fact that Passport takes less than a minute to sign up for - what documentation are you talking about? There is only one release file out there now. Also - Passport is required to *post* messages. Viewing doesn't require anything. Here's a message thread - I'm giving you a [Coral link to show that a Passport login isn't required](http://www.gotdotnet.com.nyud.net:8090/workspaces/messageboard/thread.aspx?id=ad7acff7-ab1e-4bcb-99c0-57ac5a3a9742&threadid=e7894247-670c-4e91-9319-59e691d80fa1)

Now, if you object to people having to sign-in to post on a messageboard,that's a different story all together.

Now, I had the pleasure (pain) of setting up a project on Sourceforge recently.When people can go through the complicated hellhole of a mess (I'm tempted to use much fouler language) that it takes to get a Sourceforge development system up and running on Windows (as well as the unbelievably bad quality of TortoiseCVS which is what SF recommends), no one should be complaining about GotDotNet.

And "true open source" (as defined by the OSI, I guess) is not FePy is meant to be. It is a project where
(1) You have the source code
(2) You can do anything to the code including making money from using it in a product..or forking it and creating your own project
(3) You are in no danger of violating any of MS's patents while using the source code

IMHO, I think that fits most people's definition of the word "open". if you're looking at a project that ESR or RMS would approve, you're looking at the wrong place. If you're looking for an interesting project with source code access, this is what you want.

(c) You have two questions that were asked less than 2 days ago.And it is a weekend - so give Jim a break.But let's look at those questions
- The 'read on' one. Doing a quick [Google](http://www.google.com/search?hl=en&q=+%22patents+that+read+on%22+OR+%22patent+claims+that+read+on%22&btnG=Google+Search),
this seems to be pretty commonplace in licensing agreements. My guess is that 'reads on' is lawyer speak for 'applies to'. And seeing that the FAQ says that the patents part of the license is similar to the Mozilla License and the CPL,I have a feeling that I may be right about this.But let Jim respond.

- I read through the other 'concern' but was hardly able to find one.The only thing Paolo wants is some point to be clarified.And from the tone of his mail, he doesn't seem to consider it a big deal. Quoting him "It's fine (though likely redundant)
to require that the license should not be changed and included in the sources.
It should just make it clear that distributing modified sources with a compatible
license for the changes is ok." Again - let Jim respond to this.

d) The community aspect. Hmm..this is a toughie - but looking at WiX or Wtl or FlexWiki,I see a *lot* of community activity. For Rotor, which has a much more restrictive license, there are just tons and tons of projects out there.And looking at the traffic on the FePy list as well as the number of bugs being submitted to the GotDotNet workspace (sadly..6 on the first day itself) - it would be fair to say that this community is far from dead.

Swaroop C H says:

Let the flames begin!

a) He doesn't want patches because of legal reasons, *not* because of code flux. See [this mail where he explicitly says that](http://listserver.dreamhost.com/pipermail/users-ironpython.com/2005-March/000326.html). Of course, MS is doing this to cover themselves. I have **no** objection to that. My rant is that since nobody is able to contribute, it's basically killing any community around it.

b) Requiring a .NET passport just to view documentation (let alone bug reports) is a bad thing for an *true open source roject*. I stress on the words 'true open source project'. Besides, using a crappy forum instead of a mailing list is a no-go for me.

c) There are many [unanswered](http://listserver.dreamhost.com/pipermail/users-ironpython.com/2005-March/000327.html) [questions](http://listserver.dreamhost.com/pipermail/users-ironpython.com/2005-March/000328.html) regarding the license. It just scares me off.

d) Being OSI-certified is just one aspect of it. It is the sense of having a community around a project that makes it a true 'transparent and interactive project'.

Sorry, but all these unanswered questions have made me to forget about IronPython. I used to think it was the greatest thing since sliced bread but I'm afraid to say it's toast now.

To reiterate, IronPython is a great piece of software but it is no longer a true open source project. It is a one-man show whereas a community could have helped him. That is what I have on my mind and what I am talking about.

Sriram says:

Yikes. That's a long comment! Sorry about all the formatting problems and the typos - stuggled with the small textarea

Swaroop C H says:

a) Basically, you are saying that it is not a true open source project, just a project with the source code open. I think that answers all my questions.

b) I went to the link that Jim gave in his mail, clicked on Documentation and it asked for Passport login.

c) It is sad that legal concerns are taking precedence over the actual project but it is a concern now since people want to help and they want to know what they are getting themselves into. I hope Jim helps us understand this Shared Source license better. You have to understand that nobody has used this license before as well.

d) As I said, I have come to understand that I shouldn't consider it to be a true open project. Still, I consider it a good thing that IronPython *is* source open.

e) I can't find a ChangeLog anywhere including the archive I downloaded. The examples are looking encouraging, though.

Swaroop C H says:

About the 'Subscribe to comments', I had to remove it temporarily because I am preparing to upgrade to WP1.5... I ended up waiting for the soon-to-be-released WP 1.5.1, sorry about that. I'll put it back as soon as the upgrade is done.

Sriram says:

I don't get it - people seem to be more worried about GotDotNet and Passport than actually talking about the 0.7 release and how much of a big jump it is over 0.6. If people are not looking at FePy over license and Passport wranglings, they're missing out IMHO. And since Miguel says that Mono will soon be able to run FePy - you should be able to run it on Linux as well.

Sriram says:

Hope they remove the Passport sign in for the 'docs'.There's nothing there now - just the license and the readme. As for the changelog, I would rather prefer that Jim writes up a blog post or something on the architecture for 0.7 - as it has fundamentally changed from 0.6. It's almost like a rewrite.

Raj Shekhar says:

I know quite a few interesting projects being done in Python, can you give me some thing being done in IronPython ?

Restricting access to cvs commit to just the original author does not automatically kills off the community. The best example that I can give is MySQL. No one outside of the company can make cvs commits. Yet it has a thriving community,

-- The guy in 3C-22 in Y! Blr :-)

Grant says:

Since when was IronPython open-source to begin with?

From the get-go, Jim basically said he wasn't going to release anything unless a company paid him to work on it. 0.6 had a pretty crazy license which may or may not have been OSI approved. And if it was, someone could have forked it in the past year. But no-one did.

So once again, was IronPython ever really open-source to begin with?

Swaroop C H says:

Grant, here is an excerpt [from a mail by Jim himself](http://listserver.dreamhost.com/pipermail/users-ironpython.com/2004-August/000092.html):

> There've been many good questions about the future of IronPython development
> and about how people can most effectively contribute to further development.
> I want to thank everyone who's offered to help. I've sometimes been
> frustrated by people who complain about open source projects not moving fast
> enough and want the lead developers to just work harder. Those people who
> don't just complain but offer to work to make things better are one of the
> best parts of a good open source community.

> I want to let you all know as much as I can today about my future
> development work on IronPython. I still expect to lead the development of
> IronPython to a stable and complete 1.0 release. One part of my decision to
> go to Microsoft is to have an employer who will fully support me in
> achieving this. I know that I could never have found the time to do this on
> my own as a hobby project.

This was 7 months ago.

David Mertz says:

I remember reading Jim's original announcement about joining MS. It was full of hand-wringing, but worked through to Jim stating how he was convinced that MS would support the project, keep it open source, etc. You could tell that he was sincere, and believed it. But at the same time, I just KNEW that MS was hiring him for the sole purpose of killing IronPython.

Nothing that MS does has been, nor ever will be, good for Free Software. They lie, cheat, steal, and keep getting away with it--even fooling well-meaning people into buying their newest "we've reformed" claims.

I think it is too early to say 100% that I was right (and Swaroop is right) on what's really going on with IronPython. But we're 90% there, I believe. And I am myself quite confident that conclusion will be born out.

Sriram says:

Raj - IronPython is way too young for anything to be done on it yet. But the possibilities are mouth-watering

Swaroop C H says:

All said and done, I do not doubt Jim's intentions but other factors are probably leading him to have such restrictions. I would hate to see that IronPython would not be embraced by many people because of these factors. We just have to face the fact that it is not an open source project, just a source-open project.

Anonymous says:

I wonder how Sriram will reply to David Mertz. Is he still searching the web to create the FUD?

Swaroop C H says:

Mr. Anonymous, sorry to say this but your comment would've been more credible if you actually used your real name instead of hiding behind an anonymous tag.

Sriram says:

David - did you take a look at FePy 0.6? There was so little of it - MS could have probably written it in a few weeks. I don't think FePy is going to be killed off. In fact, I think it's the opposite - FePy might become a full-fledged member of the MS stable of languages which would be really cool (read 'full VS.NET' support).

Also, as Grant already pointed out, FePy was never open source.And if you read Swaroop's quote from Jim's mail, he talks about 'help' from the open source community but never mentions anything about people working on the codebase itself.People 'assumed' that it would be a 'true open source' project - and obviously that hasn't come to be.

If MS wanted to 'kill' FePy, I doubt the most effective way is to do a ton of improvements and release the source code to 0.7.This thing runs around 80% faster than CPython (going by Jim's Pystone benchmarks). Even if Jim is exaggerating, those are some staggering numbers. FePy has already helped Mono - running the FePy benchmarks helped them sort out quite a few perf issues with exception handling.

I see no reason for MS to ever be 'good' for Free Software. If you read up gnu.org or fsf.org, it's not exactly as if they want a glowing future for Microsoft too.

Swaroop hit the nail on the head there. If you're looking for a 'true open source project', don't look here. If you're looking for a project where you can look at and fork the source code and use it any way you want, then this is right for you.

Mr.Anonymous - searching the web really isn't necessary when you have the archives of the FePy mailing list :). If you want to talk about my 'FUD', feel free to mail me at ksriram _at_ gmx DOT net

Swaroop C H says:

Sriram: Full VS.NET support would be good indeed, I agree.

Your second sentence is highly ambiguous. I am sorry but my interpretation of Jim's mail clearly differs from your interpretation of it. Maybe you mean to say that the open source community was naive in expecting a 'true open source project' because we should never have expected anything like that from the company that hired Jim? Well, I guess I can agree with that.

MS is never going to be ‘good’ for Free Software? I think it comes down to that, yes.

Swaroop C H says:

Update: Clarification regarding IronPython's license is at [Jason Matusow's MSDN blog](http://blogs.msdn.com/jasonmatusow/archive/2005/03/28/402940.aspx).

Swaroop C H says:

Philip: Did you read the second part of that sentence?

To steal someone's words, "It's the community, stupid!"

Philip says:

Hmm. Yahoo! hired the lead developer of libyahoo2. That didn't stop development on the project.

Feedback

There's no comment box, but please do email me or tweet me your thoughts and criticisms, and I will publish the relevant ones here.