Edward Dekkers <[EMAIL PROTECTED]> wrote: > > I said that I cannot imagine a case where "I would want all partitions > > on all disk drives to be removed during an OS install". Despite your > > claims, I still would never want all partitions on all disk drives to be > > removed during an OS install. Not for the two cases that you provided > > (#2 of which is true for me all of the time, by the way), nor for any > > case.
> But I will tell you a case where I love the fact Kickstart will kill > everything. > Unattended installs - exactly what I think Kickstart was designed for. There's no reason for Kickstart to zero the partition tables on disk drives uninvolved in the OS installation, whether the install is attended or not. > It can't be very unattended if I have to stand there and do the old > 'Are you sure', 'Are you really sure', 'Last chance now', dialogs. Whoever said it should do this? It should just leave disk drives uninvolved in the OS installation alone. > You set up > Kickstart to a system of your liking, then take it, plonk it in a PC you > want to install, press the button, and make a coffee. Yes, great. Exactly what I want. Zeroing partitions tables on disk drives uninvolved in the OS installation, however, is never what I want. > When you come back, the system should be ready to go as you wanted > it. Period. Exactly. Zeroed partition tables on partitions uninvolved in the OS installation is never what I want, however. Period. The installer needs to be able to repartition disk drives upon which the OS is going to be installed, and upon which any other filesystems are going to be placed, but none of this implies that it has any reason whatsoever to remove partitions on disk drives that the installer has been told not to install any filesystems onto. Zeroing the partition table on disk drives upon which no filesystems are being placed is always wrong. What function could it possibly serve? And even in the unlikely situation that you can come up with some degenerate scenario in which it would be advantageous to delete all partitions an all disk drives, regardless of whether any filesystems are being placed on those disk drives, you could easily implement this behavior via Kickstart's pre- or post-installation script hooks. > I build quite a few systems a week, and I'm very happy with the > automated installation stuff that's available. It means I can set up > about 4 PCs at a time instead of one at a time. I have never suggested anything that would prevent this. > If the customer want's data off it, I back it up to network, then > restore when finished. That doesn't sound very unattended to me. In fact, it sounds like a lot of work to preserve data that could very well have been easily left alone by the installer, assuming the data was on disk drives that weren't involved in the OS installation. > Same thing everybody should do. Never rely on data to be there after a > major install. No backups is a bad practice to get in to. Sure is. That doesn't mean the installer should remove data that it doesn't have to. For one thing, this then leaves you with only your backup copy. Which then means that there is a period of time where you only have one copy, which leaves you vulnerable to loss in the case your backup fails, unless you have redundant backups. No matter how you slice it, there is no good reason for an OS installer to remove data on disk drives uninvolved in the OS installation. > Hey, it MAY not suit every purpose, and obviously not in your case, but > please don't get into a 'right' or 'wrong' flame war. I'm a software engineer, and thus it is my duty to point out when software is not engineered properly. Despite what you seem to imply, there *is* often right and wrong ways to do something. If this wasn't the case, "software engineering" would be a meaningless term, and schools teaching good principles of software engineering would be teaching nonsense. > There is no right or wrong here, but mainly opinion. So, is there also no right or wrong about your claim that "there is no right or wrong here"? > Kickstart was written by people who want nothing for their efforts. Last I heard, Kickstart was written by Red Hat, which last I heard, is a publicly-traded for-profit corporation. > If you want it to behave differently, sign up to their project and > make a difference. So, unless I am willing to rewrite all the software in the world, I should never submit bug reports for any of it? > I really think it's wrong a tradesman blaming tools for a botched > job. Don't you? I think that it's wrong for tool-makers to refuse to acknowledge that their might be room for improvement in the tools they make. I say this as a took-maker myself, who takes pride in well-made, and well-designed tools. |>oug -- redhat-list mailing list unsubscribe mailto:[EMAIL PROTECTED] https://listman.redhat.com/mailman/listinfo/redhat-list
