On 01/05/2015 09:38 AM, David Henningsson wrote:
> On 2015-01-05 14:32, Stephen M. Webb wrote:
>> On 01/05/2015 05:08 AM, David Henningsson wrote:
>>> Second; if you contribute just a one or two line fix, then probably the biggest issue with the CLA is the paperwork. And
>>> as long as your contributions are tiny compared to the total project, sure. One can just give the code away.
>>
>> A one or two line fix does not actually require the CLA. Only contributions of a "substantial nature" (ie. a fix that
>> is not 10 lines or fewer or could not easily be described over the telephone to someone familiar with the code) require
>> the paperwork.
>>
>> If you're just proposing spelling fixes in the documentation or a one-liner to use the newer file name for an icon, it
>> should get accepted upstream without fuss. If you contribute such small fixes frequently, the sum total will exceed the
>> threshold and you will need a CLA on file for additional contributions. This is a judgement call on behalf of the
>> project manager, and the goal is to balance the inconvenience (to a contributor) of the CLA against the inconvenience
>> (to Canonical) of not having the CLA.
>
> I assume the lack of an exact and clear measure of whether a patch requires a CLA makes people err on the side of
> caution, requiring CLAs for where it should not be needed.
Maybe. I contributed small patches to GCC for years before taking the leap and signing my soul over to the FSF so I
could get them to accept more substantial pieces of work (and they could relicense it to third parties to earn revenue).
The fact that I had to give up my first born for the big stuff never disuaded me from doing the small stuff. I expect
that's different for others, especially now you can read on the Internet how evil Canonical's CLA is without actually
having to understand.
If fewer trivial contributions get made with the CLA than without it, more trivial contributions get made with the
trivial contribution exception to the CLA requirement than without it. It's a net benefit.
> That is a problem for both sides, i e, not only for the project manager, but also if you're contributor unwilling to
> sign the CLA, you would not write the patch in the first place if there was a risk it would not get in due to the CLA
> requirement.
If someone is unwilling to ever submit a patch because of their need for utter control over what someone does with a
copy of their code, it's unlikely they're going to submit that code no matter how small it is, whether there a CLA gets
signed or not. The condition of relicense is not changed, just the need for paperwork.
> There isn't a software you can feed with patches and it outputs yes or no depending on whether a CLA is needed or not? :-)
No. No robot has yet been invented that can replace human judgement in all cases. Maybe next year, but I'm a sceptic.
--
Stephen M. Webb <stephen.webb@canonical.com>
--
ubuntu-devel mailing list
ubuntu-devel@lists.ubuntu.com
Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel
