Hi Peff, On Fri, 12 May 2017, Jeff King wrote: > On Thu, May 11, 2017 at 03:47:33PM +0200, Johannes Schindelin wrote: > > > Git uses the config for remote/upstream information in favor of the > > previously-used .git/remotes/ and .git/branches/ for a decade now. > > > > Nothing in Git writes to these files any longer, and the most > > prominent user of .git/branches/ (Cogito) is long abandoned. > > > > It is time to start not only deprecating this feature, but *really* > > deprecating it. Like, really, *really*. For reals. > > I like the goal, and the patches themselves (both the deprecations and > the actual removal) all look good to me. Thanks! > If we want to start slow (and if Junio is right that there may be some > people who are actually using this feature, we probably should), step 0 > may be making people aware that it's on the horizon, even before we > start issuing annoying warnings to them. This thread does that to some > degree, but we could also put it in the release notes for a version. > That's not guaranteed to get people's attention, but it gives more > opportunity for them to speak up before installing a version of Git that > produces warnings whenever they fetch. Okay. I do not think there is any value in putting this into the release notes, of course, as 1) next to nobody reads them, at least not closely, and 2) people who still use .git/branches/ very clearly missed the release notes of a decade ago that .git/branches/ is not the way to go. Of course, it takes what it takes. If we spend 6+ cycles deprecating a feature that only one person may still use, for what we know, so be it. Ciao, Dscho