Fixes of copyright headers: Should we downport them?

[Langer, Christoph]

Hi,

> -----Original Message-----
> From: jdk-updates-dev <jdk-updates-dev-bounces at openjdk.java.net> On
> Behalf Of Andrew Haley
> Sent: Donnerstag, 26. März 2020 17:30
> 
> On 3/26/20 3:53 PM, Volker Simonis wrote:
> > Yes, I think you're opinion is overly pedantic :)
> >
> > But maybe you're right. In that case however, the established rule that
> > every developer has to update the copyright year in every file touched by
> > his change has to be revised as well.
> >
> > And, as already mentioned by Aleksey, it is not clear to me how to handle
> > arbitrary downports which among other things also change the copyright
> year
> > if you are right that only Oracle is allowed to do that.
> 
> Please let's not go down this rabbit hole. When we change stuff, we
> have to update the copyright date. This is especially true when we
> back-port patches from a file (c) Oracle dated, say, 2020 to one dated
> 2016. We must never take code which is copyright 2020, paste it into a
> file dated 2016, and not update the copyright in the destination
> file. It's easiest just to update the copyright on the file we're
> patching to the current date.

Hm, probably you're right with that. However, actually, for process simplicity's sake, I think we're doing quite some mistakes in that sense though.

Because currently, if a backport change happens to be the one that updated a copyright year, we'll do it correctly and also increment the copyright year in the backported file. But if we're bringing down a change that has happened over the year, that is, after some other change previously had to update the copyright year, we'll miss that and have newer code than the copyright year indicates in certain files.

I guess this process is that what Oracle does and what they laid out when owning the updates project. But do we need to adjust this?



Other than that, I see no point in rejecting backports of changes like https://bugs.openjdk.java.net/browse/JDK-8220414. All the change does is correct the formatting of the license header to Oracle's standards (the comma after the second year). I don't know whether this comma is legally important or just a formatting thing but in any case, I think it corrects something which was wrong before. So we could admit it. But, of course, such backports should rather be in Oracle's interest and they only affect legal compliance and don't touch functionality at all, so if nobody in the community does backport it, it shouldn't matter either.



After all, it all comes down to a legal discussion - and so, since I'm not a lawyer and not deeply into licensing laws, everything I say in this mail is likely to be wrong ... ��


Best regards
Christoph