Hacker News new | past | comments | ask | show | jobs | submit login

> How many other OSS projects have an exhaustive list on their website of everyone who is allowed to make contributions?

That is a way of thanking contributors.

If the question is, "How many other OSS projects require filing a CA before patches will be accepted?", then the answer is "many". viz. Many run by Oracle now (e.g. http://openjdk.java.net/contribute/), all apache projects AFAIK (http://www.apache.org/licenses/icla.txt), django (http://www.djangoproject.com/foundation/cla/faq/), etc. etc.

So...what "spirit" are you talking about here?




Well let's see, the choice of license is a starting point. The EPL is, per se, a bad license (intentionally incompatible with the GPL and with the showstopping choice-of-venue clause), but it's ubiquitous in Java land so we'll let them off.

In case you hadn't noticed, Oracle is widely considered to be an enemy of open source, so pointing out similarities with their processes isn't hugely helpful. In fact, contributor agreements are reviled by a large segment of the open source community (see the fights that go on whenever this subject comes up on LWN, one of the less flame-infested pro-OSS sites). This example is a particularly bad CA because it is effectively a full grant. The only way it could be worse would be a total transfer of copyright.

Finally, unlike any other projects I'm aware of, Clojure requires the agreement to be printed out, signed, then snail mailed to a foreign country. There is no reason for this other than to deter potential contributors. I've accepted one CA in the past - after fixing a crash in a little-exercised part of Ogre3D I figured the change was so minor that they could own the copyright on it as far as I was concerned. Would I have submitted the fix if I had to post a letter to New York? Not bloody likely.

Other projects (Android is a notable example) are widely derided for this sort of approach to open source - why does Clojure get a free pass?


Sounds like you object to a variety of particulars that you just happen to disagree with. A fair bit shy of falling short of some kind of essential "spirit".

The OpenJDK CA process was put in place a long time ago by Sun, generally well-liked in open source circles last I knew. And I guess you've got the same problems with all of Apache, for example (which, BTW, requires a signed copy of their CA to be faxed, at a minimum -- and I suppose some would complain about the faxing). Rough spot, there.

IANAL, neither are you, and we weren't in the room when Rich talked to his. Even if those things weren't true, I'm pretty sure Rich (nor anyone else) would accede to derision by instead of nonspecific, unconstructive griping.


>The OpenJDK CA process was put in place a long time ago by Sun, generally well-liked in open source circles last I knew.

Sun was never well liked outside of the Jave gated community. They - and their CA process - were used several times as examples in articles and presentations on 'how not to do open source' (eg. http://lwn.net/Articles/370157/ - 'How to destroy your community').

>And I guess you've got the same problems with all of Apache, for example (which, BTW, requires a signed copy of their CA to be faxed, at a minimum -- and I suppose some would complain about the faxing).

No, they accept e-mailing. See http://www.apache.org/licenses/icla.txt. I am indeed opposed to their CA requirement, but in this case it appears to be simply misguided rather than an attempt at creating a barrier.

>IANAL, neither are you, and we weren't in the room when Rich talked to his. Even if those things weren't true, I'm pretty sure Rich (nor anyone else) would accede to derision by instead of nonspecific, unconstructive griping.

I object very strongly to your dismissal of my position as 'nonspecific, unconstructive griping'. I'll try to restate it as plainly as possible:

No other example has been found of a project with this requirement. Given that there are a large number of both companies and non-profit organisations requiring CAs, none of which require a postal copy, it cannot be rationally argued that this is a legal requirement unless you are also willing to argue that all of those other companies are failing to show due diligence; I don't believe that argument would have any merit. Since Clojure has that requirement, the most plausible reason is that it is an attempt to erect as large a barrier as possible to participation (and the only other reason I can think up is that Rich Hickey is a paranoid of the tinfoil hat variety, but I've seen no other evidence of that). I cannot overstate enough how large a barrier this is; even those corporate OSS projects widely condemned for their failure at understanding the open source ideal (I've already mentioned Android; another example might be OpenOffice, most of whose developers recently decided enough was enough and jumped ship) don't make it this hard. The deliberate attempt to discourage community involvement is what I believe runs counter to the spirit of open source.


    it is an attempt to erect as large 
    a barrier as possible to participation
The "barrier" is only as tall as a postage stamp.


If you honestly think there's a way they could plausibly have made it harder, start by naming a project which has a higher barrier to entry.


I couldn't care less what other projects require. If I am excited enough about contributing to them then I will do so regardless of the barrier.


Re: "other example" - SQLite is a pretty good example of an OpenSource (Hell, public domain) project requiring a CA: http://www.sqlite.org/copyright.html

And, if you work for a company, they, too require a snail mail copyright release.


I'm starting to wonder if anyone's even bothering to read what I'm writing. You've pointed out yet another example of a project which does not require a postal form, supporting my position. Requiring a disclaimer for works for hire isn't even comparable.

Edit: now that I've had my morning cup of tea, I've properly read that link - SQLite doesn't even require any form of contributor agreement, so all you're pointing out is that they require a company to agree that work done on their time is free to be contributed by the employee. That isn't the same thing at all.


FYI, Free Software Foundation has always required contributors to GNU projects to assign copyright to FSF. Back in the day, you needed to send it via snail mail, but now AFAIK, FSF sends the package to you which you have to sign and return (via snailmail). Cite - http://www.gnu.org/prep/maintain/html_node/Legal-Matters.htm... http://www.dreamsongs.com/IHE/IHE-110.html


I wasn't aware that the FSF agreement needed to be physically posted, so thank you for pointing that out.

I will point out in return that their requirements have indeed caused a number of forks (some of which ended up being merged back in without copyright assignment TTBOMK) and acrimonious arguments despite giving strong legal assurances to the contributor that the FSF will only use the assignment for a few limited purposes.

The fact that it requires physical transfer of document stands though, so I'm satisfied that you have actually read what I wrote rather than dismissing it out of hand, and will hence shut up on the topic :P.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: