> Yesterday evening I downloaded the install42.iso, cd42.iso and all > *.tgz packages from the i386 snapshots directory on the > ftp.openbsd.org website. All files had a timestamp of Sept. 24. I then > ran them through MD5 to make sure they matched the expected checksum. > > This morning I performed two OpenBSD installs on two VMware machines; > one using the install42.iso image and the included *.tgz packages, and > one using cd42.iso and the individual packages (which I made available > via a local HTTP server). > > Once this was done I compared the dmesg output of both installs and > noticed that the install42.iso machine's kernel date is Sept. 13 while > the cd42.iso machine's kernel date is Sept. 24. A quick check of the > MD5s of the *.tgz packages in the install42.iso file show that they > are different from the packages on the FTP site? > > So I'm just wondering: in the i386 snaphots directory, do the *.tgz > packages in the install42.iso file typically "lag" behind the > individual packages available on the FTP site? Is the way to get the > most recent binaries (from -CURRENT) of OpenBSD to use individual > packages and *not* the install42.iso?
Building the install##.iso files with the X sets requires a bit of clever management. While I was on my recent vacation Peter Valchev was building snapshots and he did not know about the new build sequence required to do this right. We'll try to make it simpler, of course... In the next snapshots, it will be OK again.
