Philipp Thomas wrote:
[Could you PLEASE stop the silly crossposting and decide where you want to discuss this?]
On Sat, 17 Nov 2012 06:32:49 -0800, Linda Walsh <suse@tlinx.org> wrote:
OSus. doesn't support disk checks for XFS, so it's a moot point.
And XFS is of cause the only file system out there? Just because you think it's the only viable one doesn't mean that everybody else agrees with you.
??? I never said that. I'm using logic. Statement was made that initrd was needed to perform disk checks. That's a 'Universal' statement. (They didn't say "some"). Therefore, all I need to do to disprove it, is show 1 example where it is false and the reason is shot down. If they had said "some", I wouldn't have been able to so easily shoot it down, but if they said "some", then it can't be used to support the argument that initrd is *needed* so disk checks can be run. Did you understand that chain of reasoning. Many these days have problems with basic logic. Statement A says "need initrd"... Reason 1) Because 'disk check' is needed'. Disproval of Reason 1: This is shown to only be true in *some cases* so it can't be used to support the statement that initrd is "needed" -- only that it is needed in some, user chosen circumstances.
Adding /usr and /usr/share -- you can through those ideas out the |window.
There are a lot of things that seemed like a good idea in the past and then became obsolete.
I would like to see this -- the idea of moving things from /bin -> /usr/bin ... it was thought (though no one can give good reasons why) that this was a good idea. It's now proven not to be.
And this is why opensuse isn't open. These decisions are made by behind the scenes and are not open to discussion.
They are not made behind the scenes and they aren't closed to discussion. It's mostly the way you tend to do discussions that'll keep many people from discussing with you.
Yeah, I keep asking for evidence and reasons, and using logic, like above. People just hate that.
As for not gonna happen...maybe not, but repair packages to undo your damage can be developed and widely distributed.
Feel free to do so. This is open software so there's nothing stopping you.
I'm thinking about it -- one of the possibilities -- don't know how workable it would be .. using the mergefs, to place /usr/bin as the top writeable dir on top of /usr and /bin as top writeable dir on top of /usr/bin during package upgrade time. Things deleted out of /bin, will really be deleted out of /usr/bin, and things added to /usr/bin will really get added to /bin. When done installing, .. might keep the merge fs going, BUT, binaries will tend to collect in /bin (being on top of the dir-stack as the writable dir) in /usr/bin. Then when the system boots, one can add a shadow dir to one's root fs, that has links back to /bin for /usr/bin, or just mount a copy of /bin on /usr/bin. Then everything that is looking for things in /usr/bin will be fooled. Scripts that check for file existence during install of files in /bin and try to replace them with links will be acting on /usr/bin through the mergefs... Might be the easiest way -- but there's alot of gotcha's and if's there. More thorough, would be to skim through /bin for any links pointing to /usr/bin, and switch the links and their files -- same for /sbin, and then do a lib-dependency check and make sure those on on /lib64 (no link needed). Ultimately all the boot needed files could be back in /bin before the Dilbertian types realize what happened -- it will still boot off of /bin and mount /bin on /usr/bin (either as a complete replacement, or as a RO copy that's merged with the rest of the contents.).... That these same techniques could have been used to keep everything on bin in the first place and still create the appearance of all the bins in /usr/bin, is a matter of not having enough senior engineers reviewing the design. Anything other reason you think this "make work" was needed? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org