Re: Bug#1050001: Unwinding directory aliasing
On Thu, 31 Aug 2023 at 10:48, Sean Whitton <spwhitton@spwhitton.name> wrote:
>
> Hello,
>
> On Sat 26 Aug 2023 at 03:47pm +01, Luca Boccassi wrote:
>
> >> On Wed, 23 Aug 2023 at 10:35:42 +0100, Luca Boccassi wrote:
> >> > Suse [...] tried [symlink farms similar to those that Ian proposes]
> >> > for a few years, then had to backtrack and
> >> > go the way everyone else successfully went instead, which quickly wrapped up as
> >> > expected.
> >>
> >> I alluded to that in a previous mail to this thread, but I don't have
> >> first-hand knowledge of the specifics of how that went, and it sounds
> >> as though you might. Do you have references that you can point us to?
> >> (To the list or as private email for me to summarize on-list later,
> >> whichever seems more appropriate.
> >
> > https://en.opensuse.org/openSUSE:Usr_merge_preparation#initial_situation_in_openSUSE
>
> On this page it says
>
> The previous approach did not explain how to transition from
> [symlinks in /bin] to the actual directories as symlinks though.
>
> This implies that opensuse's goal was to get to the symlinked layout we
> have now, with something like Ian's preferred layout as an intermediate
> step. But then problems they encountered in trying to reach Ian's
> preferred layout as an intermediate step need not be encountered when
> trying to reach Ian's layout as a desired end state. So, I'm not sure
> we can use opensuse's experience as so straightforward a case to learn
> from.
Their goal was to get to the same state that we are in now since
bookworm - where one can assume there is no difference between bin and
usr/bin (et al), and do not need to care anymore. What they found out
is that with a symlinks-farm approach (that Ian and one or two more
people are fixated upon) you just don't have a realistic pathway to
get there, so they undid it and went the way everyone else, including
us, went.
Which means suse's experience is as real and concrete evidence as
anyone can get that the symlinks-farm approach does not work. And as
of today, despite asking many times (literally years), there is still
no concrete evidence that it _would_ work.
Unless of course you mean that having no difference between bin and
usr/bin (et al) is no longer a goal, but that would be something
completely different and unrelated to usrmerge, and as already
mentioned it's something completely different from what everyone else
has done, and from what upstreams (including systemd) support, and
would stop us from being able to provide hermetic /usr and read-only
images that use it among other things, so it seems to me it's
completely academic and moot.
Reply to: