Date: Mon, 9 Jan 2017 10:51:39 -0700 From: Adam Weinberger <adamw@adamw.org> To: marino@freebsd.org Cc: Adam Weinberger <adamw@FreeBSD.org>, ports-committers@freebsd.org, svn-ports-all@freebsd.org, svn-ports-head@freebsd.org Subject: Re: svn commit: r430981 - head/editors/vim Message-ID: <3C222198-383E-4705-BE5B-266096D34015@adamw.org> In-Reply-To: <50df5703-a6da-e25f-5f48-6501dbc80ed0@marino.st> References: <201701091703.v09H3cEE082911@repo.freebsd.org> <50df5703-a6da-e25f-5f48-6501dbc80ed0@marino.st>
next in thread | previous in thread | raw e-mail | index | archive | help
> On 9 Jan, 2017, at 10:22, John Marino <freebsd.contact@marino.st> = wrote: >=20 > On 1/9/2017 11:03, Adam Weinberger wrote: >> Author: adamw >> Date: Mon Jan 9 17:03:37 2017 >> New Revision: 430981 >> URL: https://svnweb.freebsd.org/changeset/ports/430981 >>=20 >> Log: >> Re-add MAKE_JOBS_UNSAFE. =46rom mat: >> --- scratch --- >> cp config.mk.dist auto/config.mk >> --- clean --- >> make[2]: = "/wrkdirs/usr/ports/editors/vim/work/vim-8.0.0149/src/po/Makefile" line = 4: Could not find ../auto/config.mk >> make[2]: Fatal errors encountered -- cannot continue >>=20 >> Install desktop files and icons when the GNOME, GTK2, or GTK3 knobs = are turned >> on. Requested by Kevin Zheng. PORTREVISION bump for this. >>=20 >=20 > Hi Adam, > So I looked up the commit history since this message made me curious = and this comes from 10 DEC 16: > "Patch 129 was a fix for parallel make. It builds fine for me on > FreeBSD with -j4, and on macOS with -j8, but that's the extent > of what I can test on my own. I'm removing MAKE_JOBS_UNSAFE with > this commit, but if one of you with your crazy 256-core machines > encounters build failures then please let me know!" >=20 > I've seen this kind of thing from time-to-time, where somebody like = me, after being 100% sure and, marks a port jobs unsafe usually = documenting why. Then later, somebody tries to recreate it on some = random machine, can't do it, and decides, "Hey, it must have magically = fixed itself" and removes the label. And then, of course, it's actually = still broken and the original committer often has to relabel the port = unsafe. >=20 > What should happen is that the original cause for jobs unsafety has to = be traced, and then either A) patched to fix to B) confirm concretely = that upstream has identified and fixed the problem. Without concrete = proof that a port has been fixed, IMO it should remain unsafe = indefinitely. >=20 > I don't think this is written down anywhere, but it would be nice if = it were documented in a guide, perhaps the do's/dont's for ports = committers because incorrect reversion of MAKE_JOBS_UNSAFE happens much = more often than it should. >=20 > No, it's not the worst thing in the world, but I think as a group we = can do better in this area. Reproducing jobs unsafety is not always = easy, nor is it a simple matter of -j number. >=20 > John You're totally right about this, and I can't disagree with anything you = said here. I had some confidence that the build problem was fixed (I ran = into the original problem 100% of the time, and with patch 129 it built = successfully 100% of the time for me), but, as you noted, wishful = thinking isn't concrete evidence, no matter how hard you cross your = fingers. I do urge you to produce an item for the PH. # Adam --=20 Adam Weinberger adamw@adamw.org https://www.adamw.org
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3C222198-383E-4705-BE5B-266096D34015>