Date: Tue, 26 Jan 2016 10:12:41 -0700 (MST) From: Warren Block <wblock@wonkity.com> To: "Brandon J. Wandersee" <brandon.wandersee@gmail.com> Cc: Tim Daneliuk <tundra@tundraware.com>, Terje Elde <terje@elde.net>, FreeBSD Mailing List <freebsd-questions@freebsd.org> Subject: Re: [installworld] Do We need /usr/obj Message-ID: <alpine.BSF.2.20.1601260956470.64883@wonkity.com> In-Reply-To: <86oac8fg5t.fsf@WorkBox.Home> References: <56A532AC.3050803@tundraware.com> <B352D17C-E1F5-44C6-A530-2CF37BFF09E7@elde.net> <56A53AA1.1010405@tundraware.com> <9D1058C8-6599-44B7-9C20-1A0F3DA48FE0@elde.net> <56A53EC5.3040805@tundraware.com> <86zivus1yv.fsf@WorkBox.Home> <alpine.BSF.2.20.1601251845450.3341@wonkity.com> <86oac8fg5t.fsf@WorkBox.Home>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, 26 Jan 2016, Brandon J. Wandersee wrote: > > Warren Block writes: > >> On Sun, 24 Jan 2016, Brandon J. Wandersee wrote: >>> >>> Second: placing NO_CLEAN="YES" in /etc/make.conf will greatly reduce >>> build time, though using it on anything but -RELEASE would probably be a >>> bad idea. >> >> Why? I use NO_CLEAN routinely to update between versions of 10-stable. >> If there is ever a problem, it's easy enough to delete /usr/obj (faster >> than 'make clean') and try again. >> >> Actually, this is part of a fairly involved build script that automates >> the normal process. > > Well, my reasoning was that changes between builds of the releng/* > branches are fewer, farther between, and usually trivial (from a code > management perspective), while changes between builds on development > branches might involve more drastic fundamental changes, moreso the > longer one goes without building. Since the OP wants to always minimize > build time (and prefers unattended builds), the increased risk of a > failed build probably wouldn't be worth it. If a build fails one can > always just demolish what's already been built and start over, sure, but > that's not really a favorable argument if always minimizing build time > is the ultimate goal. > > But I have to admit I've never tried it, and if it works, then I stand > corrected. It rarely has a problem for me, and typically takes only a third or a fourth of the time of a clean build. That comes at the cost of having about 5G of disk space permanently tied up in /usr/obj. An unattended build should be fine, nothing harmed if it fails. I would resist doing an unattended install.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?alpine.BSF.2.20.1601260956470.64883>