RE: 759a996d610d before AND after an upgrade

Graham Perrin <grahamperrin_at_gmail.com> wrote on
Date: Sat, 24 Feb 2024 14:44:46 UTC :

> main-n268493-759a996d610d for two boot environments, as shown at 
> <https://gist.github.com/grahamperrin/86359cb7c6d1316d75f340fb305484ed>.
> 
> I can't understand how this (sameness) might occur. Any suggestions?

pkgbase is still a non-production/early status. You have assumed something
is (just someimes?) not true. I recently discovered such a bad assumption
on my part. (More later.)

For:

1500014-d - - 1.81G 2024-02-23 01:15 main-n268493-759a996d610d
1500014-e - - 1.79G 2024-02-23 13:49 main-n268493-759a996d610d

FYI: 759a996d610d is the last commit on 2024-Feb-22 (UTC)

Thu, 22 Feb 2024
. . .
git: 759a996d610d - main - LinuxKPI: 802.11: implement a deferred RX path Bjoern A. Zeeb
(CommitDate: 2024-02-22 22:46:48 +0000)

Are your shown times UTC? Knowing the time zone might help with
identifying the relationships involved?


And then there is my bad assumption (referencing your
material, not my earlier example):

FreeBSD-src upgraded: 15.snap20240222221139 -> 15.snap20240223095347 
. . .
FreeBSD-src-sys upgraded: 15.snap20240222234040 -> 15.snap20240223095347

I had originally assumed that that the FreeBSD-src-sys content
matches the commit that "uname -apKU" and the like report.
This is not the case.

In the example that I hand analyzed, the FreeBSD-src-sys
expansion had commits from ~10 hours after the commit indicated
in the main-n??????-???????????? that uname reported.

The likes of "uname -apKU" at least report the commit that kernel
is actually based at. (Or  am I making another bad assumption?)
But identifying the source files that do not apply requires also
looking elsewhere so one can later find such text by content in
the FreeBSD-src-sys expansion.

It is worse for world: I'm not aware of any identified way to
systematically find what commit it is based on. From what I've
seen it seems to not be based on the commit the kernel reports
but is based on some later commit. But the FreeBSD-src
expansion need not be an exact match to that vintage and need
not match the FreeBSD-src-sys vintage.

(Even if FreeBSD-src was accurate by content, converting that
content back to a commit hash would be messy.)

Knowing what one has or if a fairly recent change is present
to test is currently a mess for pkgbase.

===
Mark Millard
marklmi at yahoo.com