Re:Don’t sign a CLA
by Sir_Cmpwn on Friday October 05, 2018 @10:58AM
(#18148643)
I think the FSF gets a pass here, given that the GPL has explicit protections against the negative behaviors warned about in the article.
[ Parent
| on HN ]
Wouldn't your article then more accurately be "Don't sign a CLA without reading it and making sure you agree with what it says", which reduces pretty cleanly to "Don't sign things unless you've read them and agree with them"?
Your overall point seems to be one I agree with, which is "CLAs can be (and often are) used to enable maintainers to make decisions that contributors wouldn't want to be made, and contributors should thus be careful in what they agree to", but in making it an imperative headline it's lost the nuance.
Well, I could make a CLA which asks you to remember to drink your Ovaltine, which would be pretty harmless. But the average CLA is far from harmless, so it deserves special mention to call out these CLAs as harmful.
Some CLAs are worse than others, though, and as people send me examples of supposedly benine CLAs I've been writing up addendums to add to the bottom of the article. There is indeed some nuance here, which I hope to address in that manner.
I can see your point of view, though one nitpick: the GPL doesn't prevent shenanigans any more than an MIT license. With a CLA, the owners can relicense the code as a non-open source license. With both MIT or GPL, you retain the rights to develop the previously released code.
It's the FSF's commitment (mentioned upthread) to not do that which arguably makes it safe to sign a CLA with the FSF.
But with the GPL, all future modifications to the code are required to also be GPL. You cannot make a closed-source fork of a GPL'd project.
A third party can’t, but the copyright holder can. That’s why some companies release GPL’d software with a closed commercial version.
We can generally trust the FSF not to, but copyright gives you that right.