Features GNOME Developers Want In The Linux Kernel

[Ansla]

it leaves the other leafs at the original level in a strange state of being older than the su directory they reside in.

I hope I'm making sense here. Trying one more time - does it not create errors in logic to have sib directories that are time stamped older than the parent directory.

Do the following operations: create directory A. Notice its time stamp.
Now inside it create directory B. Notice that the timestamp of A changed and is now the same as the one of B.
Also inside directory a create directory C. Notice that now A and C have the same timestamp that is more recent than that of B.

So B is now older than the directory that it resides in.

(note that you have to use ls --full-time or some other tool that also shows seconds to notice differences in timestamps less then a minute)

[Creak]

The thing that really bothered me here is the request to percolate time stamps up the directory tree. This seems like an exceedingly bad idea to me. If one leaf in a tree updates and the time stamp works it's way up the tree it leaves the other leafs at the original level in a strange state of being older than the su directory they reside in.

I hope I'm making sense here. Trying one more time - does it not create errors in logic to have sib directories that are time stamped older than the parent directory. It just blows my mind that somebody would ask for this.

It depends... If the timestamp is not in the existing "last modified" node property, but in a new "last subdir modified" one, this could work... and also take more memory during format, but not that much. And it should improve considerably the speed of trackers and sync softwares.

[wizard69]

The thing that really bothered me here is the request to percolate time stamps up the directory tree. This seems like an exceedingly bad idea to me. If one leaf in a tree updates and the time stamp works it's way up the tree it leaves the other leafs at the original level in a strange state of being older than the su directory they reside in.

I hope I'm making sense here. Trying one more time - does it not create errors in logic to have sib directories that are time stamped older than the parent directory. It just blows my mind that somebody would ask for this.

[Creak]

Just curious, where do you still have this? For me it was all solved with KMS several years ago.

I agree with jrch2k8, I also have X and VT fights when using plymouth, even though I use open source drivers (and thus KMS).
It's better than without, but it's still far from a neat solution.

(I tested on Fedora last time; maybe on Arch too after, but I'm not sure since my first experience wasn't a real success)

[Ansla]

He's probably using a blob driver so no KMS.

[curaga]

plymouth works but the problem is the start of X and VT fighting each other to death, with Wayland/KMS this won't happen and will behave like Os X or windows(not resolution change flipping mess at start)

Just curious, where do you still have this? For me it was all solved with KMS several years ago.

[Massa]

Oh forgot sake, this is a problem?
Add a good specification and be done with it!

Something in the line of the following variables:
- display backlight strength (0 means it's off, any other number means it's not off, minimum while still being enabled shouldn't be 0)
- display backlight on/off (0 means it's off, 1 it's on)
- display on/off
- has Backlight (Yes/no) (for e-ink displays, gotta cover all cases)
- holds content without energy (yes/no) (for e-ink or other such displays, needed to know regarding energy efficiency, gotta standardize all the information)
- is active matrix (yes/no) (needed to know what screen setting is the most energy efficient, gotta standardize this information too)

You seem to have this stuff figured out, I'm sure they'll take your patches on lkml.

[plonoma]

And an variable that tells if the screen has burn in problems

http://en.wikipedia.org/wiki/Screen_burn-in

- has screen burn in (Yes/no) Default without user added settings or if not provided then it should be yes.

[plonoma]

Thats the problem though... This got discussed at XDC (see: https://www.youtube.com/watch?v=VO7nU3-MGMw ) on some laptops a backlight of 0 means the -display- is totally off, on some it means the -backlight- is totally off, and on some it means that backlight is on but its set the absolute minimum it supports while still being enabled. All depends on the manufacturer of the machine and the age of the machine.

Oh forgot sake, this is a problem?
Add a good specification and be done with it!

Something in the line of the following variables:
- display backlight strength (0 means it's off, any other number means it's not off, minimum while still being enabled shouldn't be 0)
- display backlight on/off (0 means it's off, 1 it's on)
- display on/off
- has Backlight (Yes/no) (for e-ink displays, gotta cover all cases)
- holds content without energy (yes/no) (for e-ink or other such displays, needed to know regarding energy efficiency, gotta standardize all the information)
- is active matrix (yes/no) (needed to know what screen setting is the most energy efficient, gotta standardize this information too)

[jrch2k8]

Yeah, and for Plymouth to actually really work.

plymouth works but the problem is the start of X and VT fighting each other to death, with Wayland/KMS this won't happen and will behave like Os X or windows(not resolution change flipping mess at start)