Date: Tue, 25 Mar 1997 20:23:13 -0600 From: Richard Wackerbarth <rkw@dataplex.net> To: dkelly@hiwaay.net Cc: cracauer@wavehh.hanse.de, jkh@time.cdrom.COM, freebsd-current@freebsd.org Subject: Re: CVS repository pushed off the FreeBSD CD distribution... Message-ID: <l03010d03af5e346e2f5b@[208.2.87.4]> In-Reply-To: <199703250156.TAA20506@nexgen.hiwaay.net> References: Message from cracauer@wavehh.hanse.de (Martin Cracauer) of "Mon, 24 Mar 1997 14:28:55 %2B0100." <9703241328.AA04812@wavehh.hanse.de>
next in thread | previous in thread | raw e-mail | index | archive | help
At 7:56 PM -0600 3/24/97, dkelly@hiwaay.net wrote: > >> I always felt that a base CTM delta would be very valuable for a CD, >> too. After all, CTM is for use poor people who can't ftp a 70 MB >> file. > >Agreed. But wouldn't it be even simpler if the -RELEASE was sync'ed to a >CTM delta (might be already, would be nice if there was a release marker in >the CTM filename), and then all we'd need are .ctm_status files in the >appropriate places. Even cheaper than a 70M base delta. As the new "de facto" CTM-meister, let me comment. 1) I have already begun the process is ridding the world of redundant base distributions by placing the src-RELEASE tarballs in the ctm delta chain. The only missing point is getting the .ctm_status file included as a part of the distribution. I am still lobbying for that. However, there are technical reasons that are slowing the process. 2) For FTP, getting all the "src" directory of the distribution is only marginally more difficult than getting a base delta file. Since it saves 30Meg on each mirror site as well as the CD, I consider it a good change. I will still generate periodic base deltas for those who do not otherwise have direct access to a recent baseline. On a CD, I would prefer to see an a actual tree rather than the compressed source. If we can ever get an overlay system working reliabily, we would be able to simply note the changes. If you are following a particular release family, a major portion of the tree remains unchanged. 3) In recognition of the multiple possibilities for an original source, I have changed the suffix designation to be a bit more descriptive. For example, I use 100xEmpty.gz rather than 100A.gz to represent the 30Meg initialization starting from an empty src directory. 4) CVS represents a unique situation because the tree is extremely unstable. Most, if not all, of last weeks files therein are modified. As a result, static storage (like a CD) is quickly unusable. However, a compressed form of a snapshot of the tree does provide an efficient method to create a starting point upon which updates can be applied.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?l03010d03af5e346e2f5b>