Re: [python-committers] Use hg commit --user?
On Mar 6, 2011, at 2:33 PM, Antoine Pitrou wrote:
Le dimanche 06 mars 2011 à 13:19 -0500, Steve Holden a écrit :
In short, if someone isn't able to sign a contributor agreement we should ask ourselves whether it's really appropriate to incorporate their contributions into the code base.
What do you mean with "isn't able"? Surely everyone is physically and technically able to do so :) Now, if someone (such as Anatoly) actively *refuses* to sign an agreement when asked to, I agree they might not be reliable. But I don't think that's the case we're talking about.
I meant "isn't able" in the sense that the would-be contributor doesn't have rights in the code they seek to contribute.
Ok, but how do you know that, if they still sign an agreement?
There isn't much we can do about people willfully lying to us. This does not relieve us of the obligation to try and ensure that contributions are covered by a contributor agreement.
If you make checkins of other people's code you should be as certain as you can that you have the right to include it - since your contributor agreement states that you assign to the PSF the right to relicense your contributions.
I don't understand your reasoning. When I check in someone else's work, the author of the checkin is mostly someone else (I guess under the hood it may be more complicated, in French law it might be called a "composite work" or a "collective work", but let's try to ignore that). So *my* contributor agreement can't apply to the checkin since it is only valid for my own contributions.
We'll need to get advice on this: if you are adding the code to the code base then it is surely covered by your contributor agreement (assuming you are adding the "this code provided under a contributor agreement" notice as requested in the developer notes). If you aren't, then shouldn't you be?
Sorry, I don't understand your question: shouldn't I be what?
Shouldn't you be adding the notice?
As for the "this code provided under a contributor agreement" notice, I haven't seen it added in a long time (neither by me nor by anyone else). I'm not convinced it should pollute our commit messages and/or our code files (since it would end up basically anywhere, or at least that's the desired effect).
If you aren't happy that you have the rights to do that then I don't believe you should be checking in those contributions because they may threaten to encumber code we are licensing to third parties.
I'm not unhappy with it. I'm simply quite sure that an agreement I have signed cannot be forced on a third-party (the submitter and main author of the checked in code) who hasn't signed it. Legally, I mean.
Wouldn't it be easier and more straightforward to have these people sign contributor agreements even if you continue to check in their code?
Well, between mandating the signature of an agreement, and not mandating said signature, I think the easiest and most straightforward (both for them - who have to sign it -, for us - who have to check that an agreement exists -, and for the PSF - who has to gather and record said agreements) is the latter. All other things being equal, that is.
"The latter" meaning requiring contributor agreements? I hope so, but language is rarely as clear as we would wish.
No, "the latter" meant "not mandating said signature".
OK. As far as I am concerned, adding code to the repository that is not covered by a contributor agreement is a recipe for disaster, and I would like to hear what other committers think. I'm not sure how or when the committers list was dropped from our conversation, but I hope you don't mind me adding it back for that purpose.
regards Steve
On Sun, 06 Mar 2011 14:42:27 -0500, Steve Holden <steve@holdenweb.com> wrote:
OK. As far as I am concerned, adding code to the repository that is not covered by a contributor agreement is a recipe for disaster, and I would like to hear what other committers think. I'm not sure how or when the committers list was dropped from our conversation, but I hope you don't mind me adding it back for that purpose.
When we've had these discussions before, it was basically left up to our judgement what constituted "enough code" to make worrying about a contributor agreement necessary. Certainly it does not seem that bug fix commits of a few changed lines or even tens of changed lines is worth worrying about: how can fixing our own code incorporate someone else's copyrighted work? I think you'd be hard pressed to get such a claim by a judge. (Of course, I'm not a lawyer, so I'm probably wrong.)
Bigger chunks, especially features, yes. Perhaps we have not been as good about checking on agreements in those cases as we should have been. Anyone care to review the last N changesets to find out?
Has any progress been made on an electronically signable agreement and/or adding "posting a patch to this tracker means you have the right to contribute it and you do contribute it" language to the tracker?
-- R. David Murray www.bitdance.com
On Mon, Mar 7, 2011 at 8:24 AM, R. David Murray <rdmurray@bitdance.com> wrote:
Has any progress been made on an electronically signable agreement and/or adding "posting a patch to this tracker means you have the right to contribute it and you do contribute it" language to the tracker?
Note that the latter point (regarding implicit declarations of "right to submit" and actual submission when posting patches to the tracker) is something we're already effectively relying on when incorporating posted patches. The fact that such patches are almost always derivative of the Python source in the first place also significantly reduces the chance of legal hassles.
The situation with committers is different, because each of us can push stuff straight into the source tree without going via the tracker. The only way to keep our noses clean from a legal perspective at that point is to have a contributor agreement in place that covers everything we commit to the main repository.
The contributor-agreements-for-non-committers issue mainly comes up when there is a substantial piece of code that was originally written for something else, that is suggested as a patch to CPython (e.g. the locale neutral number parsing and formatting routines).
Still, Steve's right: when we commit something, *we're* the ones making the assertion that the change is small enough not to need a contributor agreement, so the commit record should reflect that.
If something seems iffy, bring it up on python-dev (preferably cc'ing VanL directly as well).
Cheers, Nick.
-- Nick Coghlan | ncoghlan@gmail.com | Brisbane, Australia
Le dimanche 06 mars 2011 à 17:24 -0500, R. David Murray a écrit :
Has any progress been made on an electronically signable agreement and/or adding "posting a patch to this tracker means you have the right to contribute it and you do contribute it" language to the tracker?
+1. That would be a very worthwhile thing for the PSF to spend time on, and could solve all practical issues with the contributor agreement. With a preference with David's latter proposition.
Regards
Antoine.
On Mon, Mar 7, 2011 at 1:48 PM, Antoine Pitrou <solipsis@pitrou.net> wrote:
Le dimanche 06 mars 2011 à 17:24 -0500, R. David Murray a écrit :
Has any progress been made on an electronically signable agreement and/or adding "posting a patch to this tracker means you have the right to contribute it and you do contribute it" language to the tracker?
+1. That would be a very worthwhile thing for the PSF to spend time on, and could solve all practical issues with the contributor agreement. With a preference with David's latter proposition.
Regards
Antoine.
I've raised this question on the board level in the past, and will do so again at the members meeting.
jesse.
participants (5)
-
Antoine Pitrou -
Jesse Noller -
Nick Coghlan -
R. David Murray -
Steve Holden