Date: Tue, 25 Nov 1997 11:43:11 -0700 From: Nate Williams <nate@mt.sri.com> To: Richard Wackerbarth <rkw@dataplex.net> Cc: Nate Williams <nate@mt.sri.com>, freebsd-stable@FreeBSD.ORG Subject: Re: Version Resolution? Message-ID: <199711251843.LAA27978@mt.sri.com> In-Reply-To: <l03110701b0a0c5b23725@[208.2.87.4]> References: <l03110700b0a0b8ad27f2@[208.2.87.4]> <l03110701b0a0a3722b14@[208.2.87.4]> <l03110707b0a090dacca7@[208.2.87.4]> <l03110703b09f8a1710e6@[208.2.87.4]> <l03110700b09e72675ae9@[208.2.87.4]> <199711240216.CAA28304@awfulhak.demon.co.uk> <199711240504.WAA22051@mt.sri.com> <199711241922.UAA21949@bitbox.follo.net> <199711242223.PAA24374@mt.sri.com> <199711251530.IAA27130@mt.sri.com> <199711251649.JAA27402@mt.sri.com> <199711251758.KAA27804@mt.sri.com> <l03110701b0a0c5b23725@[208.2.87.4]>
next in thread | previous in thread | raw e-mail | index | archive | help
> >> >The CVS bits are available to them because > >> >it was easy to do so and it wasn't too much of a burden on the > >> >developers. But, if you aren't looking at them, there is absolutely *NO* > >> >need to get them, since CVSup of the bits is more effecient and uses > >> >less space than getting the CVS tree. > >> > >> You miss my argument that the present methodology FORCES anyone who has ANY > >> need for current revisions to take ALL of them. > > > >And you miss the argument that anyone who needs to look at recent > >history can do that now without the CVS tree. > > Not if they lack the on-line access to which you refer. What reference? If you don't have on-line access, you can't be a developer. > I am looking for things to simplify the HUGE startup and continued > processing cost of repeatedly processing the same unchanging data. What processing? > As it is, I am spending many CPU hours each day just verifying that the > bulk of the cvs tree has not changed and localizing the small portion of > the files that did change. If you're using CPU hours/day, then you've proven you've got no clue. I've got a 486/66 at the wrong end of a 28.8K line at home that is the CVSup server for my entire FreeBSD 'stable' of machines at work, as well as my development boxes. It takes literally *10* minutes to update all my machines every day, and most of the bottleneck is that silly 28.8K link sending out the same bits to every machine. Yes, it could be made more effecient by using a local CVSup server on my local LAN, but since FreeBSD is a hobby I didn't feel it was appropriate to stick the CVS bits on the company machine, plus it makes things faster at home. > Further, all the CVSup servers are having > to do similar analysis passing over a lot of unchanged data for every > client update. What analysis? If you mean MD5 cksums on the data, you aren't going to make any significant change by removing 90% of the history. *Maybe* you'll decreate it by a third, but 33% faster of < .1% of the total CPU burned is insignificant. The only load that is significant on these servers it the network bandwidth of sending out data, and that 'data' load isn't going to change unless people stop making changes to FreeBSD, which is apparently your goal. In any case, you've made history by becoming the *first* person to go into my .procmailrc file. No one else has gotten that distinction, since most people have something significant to add to the discussion, but has been obvious over the last years you have no intention on ever doing anything but bitch and moan. :( Nate
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199711251843.LAA27978>