« first day (4121 days earlier)      last day (816 days later) » 

7:26 AM
I was just having a sort-of discussion with someone about upgrading Linux-based systems.
He said Debian (for example) isn't upgrade-friendly years after the event. In part because the older distributions go away.
So does Debian lose everything older than oldstable?
 
 
1 hour later…
8:50 AM
 
@StephenKitt Could one upgrade from there, though?
Say you were on Debian 6.x.
 
@FaheemMitha I’m curious why you would think one couldn’t. However if you were on Debian 6.x, I would say that you haven’t cared about upgrades for a very long time, and you’d be better off starting from scratch on Debian 11 instead of pretending that the ability to upgrade is suddenly very important.
 
9:17 AM
@StephenKitt Why I would think one couldn't? No solid reason. But just because it's archived doesn't mean it's archived in a way that permits upgrading.
@StephenKitt And yes, if one was on Debian 6, perhaps a reinstall would be better.
 
@FaheemMitha the repositories are copied wholesale, with all the information needed. Note however that you don’t need a repository containing the version you’re upgrading from (unless your setup isn’t up-to-date with that version), you need a repository containing the version you’re upgrading to.
 
@StephenKitt Yes, I'm aware of the latter.
 
9:35 AM
The only gotcha would be expired signatures but that can be worked around.
 
@StephenKitt Can't you just instruct apt to not do verification? Or is there a better way?
 
@FaheemMitha that is the way to do it.
 
@StephenKitt OK. Thank you.
 
 
2 hours later…
11:27 AM
Debian Policy 9.3.1 says:
> Packages that include system services should include systemd service units to start or stop those services.
If I understand that correctly, it means systemd units will be enforced at some later time.
Fetchmail doesn't have one, for example.
 
Packages are still allowed to ship init scripts instead:
> If the package does not include a service unit (if, for example, no one has yet written one), including an init script, as described below, to start the service is encouraged.
 
11:53 AM
@StephenKitt I realise that. I was just commenting on systemd, since it's touted as the new standard.
 
I haven't read the policy, but that quoted part doesn't sound optional or future-based to me: "packages ... should include systemd service units..."
Unless "should" is meant more as a "please" than a "do it"
 
12:10 PM
In the "you learn something new every day" department:
88
Q: What are the --start-group and --end-group command line options?

pic11What is the purpose of those command line options? Please help to decipher the meaning of the following command line: -Wl,--start-group -lmy_lib -lyour_lib -lhis_lib -Wl,--end-group -ltheir_lib Apparently it has something to do with linking, but the GNU manual is quiet what exactly grouping me...

I had no idea such a thing existed. Not sure if using it is a good idea or not, though.
@JeffSchaller "should" has a specific meaning in Policy. It basically means it will be enforced "later". For some value of "later".
It may have a more precise definition, but I forget.
And enforced means that if you don't do it, it counts as a release-critical bug.
 
12:27 PM
@FaheemMitha I don’t think there’s any notion of “later enforcement”; see debian.org/doc/debian-policy/ch-scope.html:
> The terms should and should not, and the adjective recommended, denote best practices. Non-conformance with these guidelines will generally be considered a bug, but will not necessarily render a package unsuitable for distribution. These statements correspond to bug severities of important, normal, and minor. They are collectively called Policy recommendations.
What constitutes a release-critical bug isn’t determined by Policy, it’s determine by the release team; see release.debian.org/testing/rc_policy.txt
 
@StephenKitt My mistake. I thought that things gradually transitioned from "should" to "must".
 

« first day (4121 days earlier)      last day (816 days later) »