Description of problem: Format early, right after setting mount points, so people don't waste so much time if format fails. That's how Mandrake does it. Version-Release number of selected component (if applicable): 9 How reproducible: Didn't try Steps to reproduce: 1.Attempt dd graphical NFS install 2.Spend considerable time on package selection 3.After format failure, see alert saying press <return> to reboot 4.Click on OK button in alert Actual results: unexpected reboot Expected results: Asked me where to save package selection list before rebooting. Additional info: Except for formatting, I fully prepared /dev/sdc prior to booting from bootdisk.img for an NFS install. Since this was my first RHL install since 7.3, I spent considerable time in package selection, probably close to two hours. Only after spending so much time did the installer attempt to format the selected partitions. I had chosen to scan for bad blocks on /, so formatting also took considerable additional time at this point. When 100% was reached, an alert came up to say format failed due to excessive bad sectors, telling me to press <enter> to reboot. I didn't want to reboot, as I expected some means to save a package selection list. I clicked OK on the alert, and the system rebooted immediately. After this failed install, I immediately ran IBM's DFT program three times successfully on /dev/sdc, which means format should not have failed. 53c875 SCSI no ATA K6/2-550 CPU 256MB PC100 SDRAM SiS530 Chipset Realtek 8139 NIC Yamaha 744 Sound Brand new install target device /dev/sdc IBM DPSS Expected results really should be yet another enhancement request bug.
This is difficult to do without requiring more RAM. With the current installer framework, I'm not planning on changing this around.