2.(year-2000).x [LWN.net]

2.(year-2000).x

Posted May 24, 2011 10:04 UTC (Tue) by przemoc (guest, #67594)
Parent article: 2.8.0?

There are various ideas how to improve version numbering in Linux.

{2.8,3.0}.x, year.month, etc.

I have other proposal, that does not address particularly Linus' concern, yet still has some upsides.

2.(year - 2000).(continuity of current patch version)

I wouldn't introduce it before 2012, to avoid for the last time use of odd number and to prepare users for upcoming change.
Presumably first release in the new year would be:

2.12.42

and later 2.12.43, 2.12.44, 2.12.45, 2.13.46, and so on...

It's good because of:
- 2.x preservation,
- hinting the year of non-stable-point release,
- continuity in patch (AKA Makefile's sublevel) version numbering, which is presumably the most important part here - so you can use it alone [e.g. .42] w/o ambiguities.

And in year following .99 hit (in ~14 years), patch version would be zeroed to avoid use of 3 digits (obviously for releases after .99 in the same year it is unavoidable). I really doubt that in future anyone will be using 14 year old kernel, so such resetting isn't harmful at all.

YMMV

2.(year-2000).x

Posted May 24, 2011 13:35 UTC (Tue) by kragil (guest, #34373) [Link] (1 responses)

How the hell is this going to silence the voices in Linus' head??

Read the PS. again.

2.(year-2000).x

Posted May 25, 2011 19:01 UTC (Wed) by przemoc (guest, #67594) [Link]

Why it should? Linus head is Linus' and his family's worry. I don't have right to inject nor eject anything from there. I may only stick some ideas via discussion, but that's all.

Maybe read again my comment, where I clearly wrote that "[my version numbering proposal] does not address particularly Linus' concern".