Date: Sun, 29 Mar 2020 15:33:35 -0600 From: Warner Losh <imp@bsdimp.com> To: Ed Maste <emaste@freebsd.org> Cc: =?UTF-8?Q?Ulrich_Sp=C3=B6rlein?= <uspoerlein@gmail.com>, git@freebsd.org Subject: Re: Strange commit Message-ID: <CANCZdfogy1XSqHwbvRjArJ7X1uE4ySvye%2B82UO%2BboVWzJFP%2B7w@mail.gmail.com> In-Reply-To: <CAPyFy2AG8dDwRzEYvRKUB7eL=fPFWYrte=y1-wwLjz=HF%2BM_8Q@mail.gmail.com> References: <CANCZdfrRuLf6Q3Q-XzDe5wpSV3XCvkhht2ESh=bXbL6wAtt87g@mail.gmail.com> <CAPyFy2Dm12BD4yuGo65T_tQrGXapZxvSSurO=Gwn=2S1drq6ig@mail.gmail.com> <CAJ9axoRs9zWXteeDmZ8L1nucBPqqB-WMGuXTRFyVuxo2h4F11w@mail.gmail.com> <CAPyFy2BgeP=WxeZeYVS3z%2BgzArJe1h9ofO42O7HvznpLBsBKbg@mail.gmail.com> <CANCZdfo8m4Kg%2BXTbUJYwi1gxOr8C9RobBR0DcZ6D6veiKkOB2Q@mail.gmail.com> <CAPyFy2DROg_v47QrxmK-1TuFzqTqf=2e63N7qoX1_V=GyDkyRg@mail.gmail.com> <CANCZdfpoROXZOn0CdX6w=TK03FWc-f8sR9XJQ4cxb_Mq9GBUqw@mail.gmail.com> <CAPyFy2BcEgpuK3JDa6AUGsoGxwKmY1ozk-HqRNarvQMGtXFpwg@mail.gmail.com> <CANCZdfo1d5dLkbAAnOVYLOPwbdP-FyTMJ7T%2BDYvzabYr0vfQ8g@mail.gmail.com> <CAPyFy2AG8dDwRzEYvRKUB7eL=fPFWYrte=y1-wwLjz=HF%2BM_8Q@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, Mar 27, 2020 at 3:21 PM Ed Maste <emaste@freebsd.org> wrote: > On Fri, 27 Mar 2020 at 17:06, Warner Losh <imp@bsdimp.com> wrote: > > > > I'm sorry, that's a *HUGE* change from what we decided. And it doesn't > lose the ability for us to maintain our own local changes easily... > > My understanding has always been that we would maintain the vendor > history in our tree, as we do today in the existing svn2git conversion > (i.e. https://github.com/freebsd/freebsd) with elftoolchain. This is > the configuration that I have prototyped and experimented with. > > This does not and has never implied that we'd have *all* of the > upstream history of every piece of contrib software that we use, just > that in the history for a contrib component we have our versions > (i.e., what's in contrib/ today) as well as the snapshots > corresponding to upstream revisions (i.e., vendor/). > Ah yes. You're right. I'd gotten confused as to the exact context. Absent the ability to automatically and retrospectively generate git subtree/subrepo commit metadata (which would be quite hard and likely error-prone), that's the only way to record that. Warner
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CANCZdfogy1XSqHwbvRjArJ7X1uE4ySvye%2B82UO%2BboVWzJFP%2B7w>