Linux 3.0?

Posted Sep 4, 2008 15:15 UTC (Thu) by jengelh (guest, #33263)
In reply to: Linux 3.0? by stijn
Parent article: Linux 3.0?

>The idea of 2008 in a version number does not appeal to me, those first two bytes are really wasted.

Please thought the same in the 20th century and used two-digit year numbers everywhere (like 24.12.21 to denote 1921-Dec-24), and that backfired when it approached the year 2000. Truncating a year number to two-digits is like retrieving the short SHA for a git commitit only works at this point in time. The next commit may cause the length of the shortest possible unique SHA to increase, which is why SHAs in commits are often not abbreviated at all, just to *keep* it unambiguous for the future.

Linux 3.0?

Posted Sep 4, 2008 15:30 UTC (Thu) by stijn (subscriber, #570) [Link] (1 responses)

Whoever said anything about truncating? I've just chosen to arbitrarily set the starting point to 2000 -- it is a new epoch. I am pretty sure I'll never make releases in the year 507. If somehow people decide to stick my version tag in a 6-byte field, that is not my problem. Come the year 2100, I'll happily release (very happily I should say, if still around and able) version 100-152.

Linux 3.0?

Posted Sep 4, 2008 19:33 UTC (Thu) by dlang (guest, #313) [Link]

well, then why not use 1900 as the epoc (date already does this and returns 108 for this year)

it ends up being confusing.

and you are always free to truncate the version number yourself. think of vehicle model years, you refer to the 2008 model as the oh-eight model and everyone knows what you are referring to, but if you referred to things as the 8 model most people would take a few seconds to figure out what you are talking about.

Linux 3.0 - Date-based release numbers

Posted Sep 6, 2008 0:45 UTC (Sat) by giraffedata (guest, #1954) [Link]

I think two-digit years make sense now just like they did in 1980. The cost of carrying those extra digits all that time exceeds the cost of dealing with the century turnover. Think of all the systems that didn't even survive until 2000; 4 digit years would have been a total waste in them. What's the probability that the Linux kernel will still be around, released in the same way as it is today, in 2100?

I wrote programs in 1995 that could not survive the Y2K transition. Some had to be restarted after the turnover and others had to have minor code changes after the turnover, with minor work stoppage until that happened. Many were no longer in use. It was a net win.