New AMD X2 64-bit machine, trying to install games-fps/doom3-data-1.1.1282-r1 (as a dependency for games-fps/doom3 which is what I *really* want). I have the boxed-retail version (ISBN 1-58416-543-X), with three CDs. At root console, I set 'export CD_ROOT=/mnt/cdrom' and for good measure I also add 'export CD1_ROOT=/mnt/cdrom', 'export CD2_ROOT=/mnt/cdrom', 'export CD3_ROOT=/mnt/cdrom'. 'printenv' verifies that all are set. /etc/fstab contains the line: /dev/cdrom /mnt/cdrom auto auto,users,ro 0 0 Reproducible: Always Steps to Reproduce: 1. in root console: 'export CD_ROOT=/mnt/cdrom' and other similar env-variables noted above. 2. insert CD, verify successful mounting at /mnt/cdrom 3. initiate installation: 'emerge -v doom3-data' Actual Results: The ebuild starts fine, displays license, then finds and copies files from Disk 1. After some minutes, the ebuild then attempts to copy files from Disk 2 (which is still the same physical media as Disk 1 - no disc-change made/possible/requested) and aborts immediately, with "install: cannot stat `/mnt/cdrom/Setup/Data/base/pak000.pk4'". Expected Results: The ebuild should proceed through all the named Disks which it is expecting, and successfully install. If/when a disk-change is necessary, the ebuild should gracefully pause and elicit new media insertion by the user - maybe even helpfully eject the prior disc. The mounted disk contains these files: Path: /mnt/cdrom/Setup/Data/base/ game00.pk4 pak002.pk4 You can see that the requested/failed file (above) does not exist on this Disk. Umount/eject this Disk 1, insert/mount Disk 2, and I see: Path: /mnt/cdrom/Setup/Data/base/ pak000.pk4 pak001.pk4 My optical drive is an LG GH20LS10, which is a SATA device. Appears to be using kernel driver 'ahci' and kernel modules 'ata_generic, pata_acpi'. Kernel is 2.6.25-gentoo-r6 #1 SMP PREEMPT Wed Jul 9 17:43:13 PDT 2008 x86_64 AMD Athlon(tm) 64 X2 Dual Core Processor 6400+ AuthenticAMD GNU/Linux. UT2004 did install fine, from a single DVD, and the ebuild called for successive CD's (which were contained on the single DVD) - this ut2004 ebuild proceeded just fine.
clearing CD_ROOT appears to solve this problem: 'unset CD_ROOT'. Other variables noted in original report are left unchanged (but, I note, they are also incorrectly/unhelpfully set :-( ). With CD_ROOT unset, the ebuild proceeds to normal completion, after pausing for disc-changes and using all three retail CDs.
From eutils.eclass: # now we see if the user gave use CD_ROOT ... # if they did, let's just believe them that it's correct Looks like this is intended behaviour. You shouldn't need to set CD_ROOT unless you want to install from somewhere other than a CD mount point.