Bug#704424: release-notes: sufficient-space hints give confusing advice
Justin B Rye wrote:
> Justin B Rye wrote:
>> [...] It seems to me we should put all our coverage of
>> redundant packages in 4.4.3 (mentioning "apt-get autoremove" and
>> avoiding the word "obsolete"), put all our coverage of relic packages
>> in 4.9, and leave the two sections completely unconnected.
>
> Here's an attempt at a patch to do that.
I haven't tried to format it, but it certainly has more of a chance of
being well-formed than the patch I sent before. ;-)
So for what it's worth, this gets my ack.
[...]
> Is it still true that deborphan is "highly recommended", or is that a
> leftover from the days when it was the only tool that implemented any
> of this functionality?
I wouldn't mind a change that stops recommending deborphan.
> And does popcon-largest-unused still work? It thinks I don't use my
> web browser or window manager. Is it trying to find unused packages
> by looking at the file access datestamps on a /usr file system that's
> mounted noatime? Yup, and even relatime (now the default) will cause
> trouble - see #298760.
I think it should work with relatime, since that just ratelimits atime
updates to once per day (unless mtime or ctime < atime). The "mount"
manpage is seriously misleading. (Reference:
linux/fs/inode.c::relatime_need_update())
[...]
> Except that once I've collected under one bulletpoint all the ways of
> finding removal candidates on the basis of redundant ex-dependencies,
> it becomes obvious that popcon-largest-unused doesn't fit there; it
> fits with all the other suggestions for finding removal candidates on
> the basis of size, which are currently split between several
> bulletpoints. So I've consolidated all of those, too.
Looks good. Thanks much.
Jonathan
Reply to: