Date: Tue, 25 Dec 2001 23:30:38 -0500 (EST) From: Robert Watson <rwatson@freebsd.org> To: Jordan Hubbard <jkh@winston.freebsd.org> Cc: The Anarcat <anarcat@anarcat.dyndns.org>, re@freebsd.org, qa@freebsd.org, jkh@freebsd.org, phk@freebsd.org Subject: Re: regression tests for possible sysctl kern.disks/libdisk changes/mfc Message-ID: <Pine.NEB.3.96L.1011225232951.39578J-100000@fledge.watson.org> In-Reply-To: <40166.1009249432@winston.freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Sounds good to me. I just wanted to make sure there were no more serious failure modes possible and/or likely, since I wasn't sure of the entire rationale for the original change. I've already issued an MFC approval for this change, and hopefully it will be included in the first snapshot in a week or two. Robert N M Watson FreeBSD Core Team, TrustedBSD Project robert@fledge.watson.org NAI Labs, Safeport Network Services On Mon, 24 Dec 2001, Jordan Hubbard wrote: > 1. Nothing "broke", it simply confused the users significantly > that "da0" used to be the very first item in the disk selection > menu and then suddenly got shuffled to a different location in > that menu, rendering their muscle memory for that menu obsolete. > > 2. The downsides are mostly that the code in libdisk doesn't get > to benefit from a cleaner way of getting disk info (the sysctl). > > 3. The risks I can see is that the menu goes back to exhibiting the > funky behavior of before, but that's probably unlikely. > > - Jordan > > > > Jordan, > > > > Can you clarify a couple of things for me? > > > > (1) What exactly broke in sysinstall due to not having the names sorted? > > Was it just a slight visual glitch, or were there fairly material > > problems? > > > > (2) What are the downsides to not merging this feature: do we lose the > > ability to do something that we might otherwise have? Do things look > > better? > > > > (3) What are the risks? What can possibly go wrong? (sysinstall might > > multiply list devices, might not list some devices, might offer > > devices inappropriate to install on, ... > > > > I'm cerainly not opposed to this type of change, but if there are > > observable risks with no real payoff, I'm not sure it's worth those risks. > > If there's little risk, and lots of payoff, then that is different, of > > course :-). > > > > Robert N M Watson FreeBSD Core Team, TrustedBSD Project > > robert@fledge.watson.org NAI Labs, Safeport Network Services > > > > On Sun, 23 Dec 2001, The Anarcat wrote: > > > > > Hi. > > > > > > A recent commit to -current was done to re-enable the use of sysctl > > > kern.disks by sorting its output in libdisk. > > > > > > This might break sysinstall as it is why the disabling was done in the > > > first place. > > > > > > I have mildly tested sysinstall with the -stable patches, and it seems > > > to order the disks properly. I haven't tested an install or real-life > > > partitioning. > > > > > > Those wanting to do so need to apply the attached patch and rebuild > > > libdisk + sysinstall: > > > > > > # cd /usr/src/lib/libdisk > > > # patch < patch > > > # make clean all install > > > # cd /usr/src/release/sysinstall > > > # make clean all install > > > # /stand/sysinstall > > > > > > Since I am not aware of what precisely the original bugs were, I cannot > > > fully test this. > > > > > > Also, I won't be available to test this in the following few days. I > > > will be back on the 27th and might take my mail sporadically in between. > > > I will stay for a few days and leave again until the 2nd, so I will be > > > available for real testing only then. > > > > > > Thanks for your attention, > > > > > > A. > > > > > > > > > > To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-qa" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.NEB.3.96L.1011225232951.39578J-100000>