Date: Mon, 01 Sep 2014 17:01:15 +0200 From: John Marino <freebsd.contact@marino.st> To: Adam Weinberger <adamw@adamw.org>, Alexey Dokuchaev <danfe@FreeBSD.org> Cc: svn-ports-head@freebsd.org, svn-ports-all@freebsd.org, ports-committers@freebsd.org, Pietro Cerutti <gahr@FreeBSD.org> Subject: Re: svn commit: r366841 - head/lang/tcl86/files Message-ID: <54048A3B.4030001@marino.st> In-Reply-To: <65B530D9-4740-4A60-A2F5-40335A520C4E@adamw.org> References: <201409010731.s817Vrxf062753@svn.freebsd.org> <20140901074609.GA32100@FreeBSD.org> <65B530D9-4740-4A60-A2F5-40335A520C4E@adamw.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On 9/1/2014 16:47, Adam Weinberger wrote: > On 1 Sep, 2014, at 3:46, Alexey Dokuchaev <danfe@FreeBSD.org> wrote: >> Since when did we abandon the practice of doing 'svn diff' before >> commit? > > Alexey, please direct your frustration elsewhere. Pietro fixed > (hopefully) a problem that caused a lot of tcl-dependent ports to > break. He should be commended, not sneered at. If you don’t like the > way makepatch works, take it up with portmgr, not with the committers > who use it. I know for a fact that portmgr is interested in extending > makepatch to DTRT. > > Besides, there are simply situations where makepatch is just not the > tool to use. Sometimes I *have* to run diff(1) by hand. Sometimes I > need to run "make patch” before I can edit the diff. Sometimes I’ve > edited multiple files and I want to test one diff at a time. > Sometimes a file needs to be touched by two different patches (I > don’t like it, but ten years ago it seems like people really did). While I also pushed back against danfe@ on this post, it's obvious that makepatch was used. The commit didn't do anything except change the timestamp. Danfe *did* rework makepatch, I helped him, and it's stuck in review. His proposed version would not have resulted in a commit that just changed timestamps. It does solve this problem. > If you REALLY want to solve that issue, make a change to > src/gnu/usr.bin/diff so that it will take a flag to not put > timestamps in there. As I said, the issue has been solved, and the solution is good. Nobody dislikes the new changes internally*, but patch-naming has turned into a impasse. John * since today, antoine says he thinks -p option on diff is "ugly". it's a highly useful option so now we have yet another hurdle to jump. If not for phabric we could have had this in ports weeks ago, but now are stuck in an impasse (which I suspect was the outcome desired by the people that wanted it reviewed in phabric tbh)
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?54048A3B.4030001>