| 1 |
On 09/29/2013 09:03 PM, Greg Woodbury wrote: |
| 2 |
> One of the most obvious things that broke booting with a seperate /usr |
| 3 |
> is not GNOMEs fault, but GRUB 2's fault. |
| 4 |
> |
| 5 |
> the move of /bin to /usr/bin (for things like cp,mv,ln,ls) came after |
| 6 |
> the breakage of /usr, but is symptomatic of some distros cavalier |
| 7 |
> attitudes to the problem. |
| 8 |
> |
| 9 |
> /usr/lib/udev..... |
| 10 |
> /usr/lib/systemd..... |
| 11 |
> |
| 12 |
> were both placed in /usr despite objections from a number of folks. |
| 13 |
> |
| 14 |
> So claims that udev and systemd are not responsible are not true. |
| 15 |
> |
| 16 |
> /usr/lib/e2initrd-helper was placed in /usr despite objections. |
| 17 |
> /usr/lib/libdevmapper* was moved despite objections... |
| 18 |
> /usr/lib/liblvm2* helpers were placed in /usr despite objections... |
| 19 |
> |
| 20 |
> There were deliberate placements of "new" or updated libraries in /usr |
| 21 |
> that were known ahead of time that would break the use of a separate |
| 22 |
> /usr filesystem. |
| 23 |
> |
| 24 |
> It was pointed out quite plainly at the time, and the placements were |
| 25 |
> made anyway, dismissing the concerns are "mere historical artifacts" or |
| 26 |
> "clinging to ancient and outmoded traditions." |
| 27 |
> |
| 28 |
> The *same things* are still being cited (about being outmoded) in |
| 29 |
> dismissing concers about forcing useres to adopt technologies they do |
| 30 |
> not want to use. |
| 31 |
> |
| 32 |
I agree with you but I think the distinction made on udev and systemd is |
| 33 |
that their *upstreams* didn't suggest they go in /usr. I don't know for |
| 34 |
sure since I don't follow systemd and frankly detest it, but if they |
| 35 |
said systemd and udev belong in /bin or somesuch, then they can't be |
| 36 |
held responsible. The responsibility for that decision falls on the |
| 37 |
maintainers, but it's compounded by the dependencies of systemd/udev. |
| 38 |
They depend on dbus, which resides in /usr. Was that an upstream |
| 39 |
decision or a distro decision? If we follow this dependency tree of /usr |
| 40 |
moves, which package was the first to migrate to /usr and begin this |
| 41 |
problem? The move to /usr was a social one that started years ago, not a |
| 42 |
technical decision made this year. The issue right now is damage control |
| 43 |
and, for some, maybe taking a good look at the FHS and deciding what |
| 44 |
really makes sense. |
Archives