Re: [H-source-users] Shall we remove debian from h-node?

[Denis 'GNUtoo' Carikli]

On Sat, 6 Jul 2024 10:52:34 -0400
bill-auger <bill-auger@peers.community> wrote:

> ok, so this is a summary, now that denian 12 is released and we have
> seen it's form - it is most likely that all newly installed debian
> systems would be highly prone to false-positives - most hardware will
> "just work" and there is no easy way for the user to know if it works
> only because of non-free firmware - the short answer is "yes"; but we
> have suspected and later known that for some time already - since
> this was last discussed, RMS has also concurred that debian 12 should
> not be an option for h-node
For me it's obvious that new entries with "Debian 12" don't make any
sense, the question was always what to do with older entries.

Since users are not going to install RC of Debian 12 anymore, we don't
need newer tests with Debian 12.

Instead the information we want is "does this hardware works without
nonfree firmwares", and "how can I reproduce this test", and the
information we don't want is "does this computer run on Debian 12
because as a user I want to run Debian 12".

So far I found in the xml dump, in the distribution field:
- debian
- debian_12
- debian_testing
- debian_unstable

And we also have things like "Debian 12", "Debian stable", etc in the
text.

So I propose to try to replace debian, debian_testing, debian_unstable
or even "Debian stable" to something with a version number. We could
have debian_11_unstable, or simply debian_<number smaller than 12> when
other information enable to deduce that.

For the entries that also have distributions that are known to work
(like debian_11, trisquel_<version>, etc), we could probably remove the
mention of the problematic Debians completely (either debian_12 or
some unknown version of Debian) as it brings no additional useful
information: we already can reproduce on other systems with Trisquel
for instance.

Remains what to do with entries with only bad Debians like debian_12
(we have some). And here some rename of debian_12 can help: If we have
some way of confirming that the user didn't use nonfree firmwares we
could rename the 'debian_12' to something like
'somewhere_between_debian_11_and_12_without_nonfree_firmwares', and do
the same for other debians without a version
('somewhere_between_debian_unstable_and_12_without_nonfree_firmwares').

If that can help we could also have some bogus name like
suspicious_debian_12 to tag what needs to be looked at.

> really, debian 12 should not have been added in the first place - it
> was added 2 years before release
The best way to deal with it would have been to characterize the data
better: to have debian_12_rc1 and keep up with the rcs, or just not add
debian_12 as you suggested, especially since adding in nonfree
firmwares was being discussed.

> so this is not as straight forward as i could have been - it will
> need some sorting out - i have recently been granted admin access;
> and i am looking into how to address these urgent concerns now
As I understand any user can also edit any entries. Though maybe as
admin you also have extra information like when an entry was submitted.
This information could help replacing "Debian stable" by the correct
version number for instance. An idea could also be to contact users
that added "debian_12" entries that we can't easily check.

In addition we probably need to warn users about debian_12 and similar
to explain the situation for all perspectives (people wanting to use
the data, people contributing data, etc).

Denis.

pgpmwkAUrTebD.pgp
Description: OpenPGP digital signature