On: splice() and the ghost of set_fs()

[theteapot]

Shouldn't they have left it for 6.0?

[mhitza → theteapot]

Kernel doesn't adhere to a SemVer policy

[theteapot → mhitza]

Maybe they should. Maybe, they, should.

[SAI_Peregrinus → theteapot]

Naw. Kernel version number increases. Can't be more than 255 in any field, except the EXTRAVERSION which is a string. Kernel version numbering is exposed in the userspace ABI, so changing it to have meaning would be a breaking change for no substantial benefit. This[1] LWN article is a good overview of the history. Effectively the major gets incremented when the minor gets to 20 and Linus runs out of fingers & toes to keep track.

[1] https://lwn.net/Articles/871989/

[mrlonglong → SAI_Peregrinus]

How many digits does a penguin have?

[ufo → mrlonglong]

I believe it's 14. (Birds feet have 4 toes and their wings have 3 fingers)

[theteapot → SAI_Peregrinus]

> Can't be more than 255 in any field, except the EXTRAVERSION which is a string. Kernel version numbering is exposed in the userspace ABI, so changing it to have meaning would be a breaking change.

Sounds like they should leave this change for 6.0 then.

[SAI_Peregrinus → theteapot]

Why? The kernel version number has no meaning, other than to provide an ordering of releases.

[theteapot → SAI_Peregrinus]

In 6.0 it will.

[charcircuit → SAI_Peregrinus]

>Can't be more than 255 in any field

They can just change the max. It's not that big of a deal.

[SAI_Peregrinus → charcircuit]

No, it's one byte per field (except EXTRAVERSION, which is a string). That's part of the userspace ABI, so it can't change without extremely good reason (critical security issue that can't be mitigated in another way).