Guix role in a free society

Guix role in a free society

[Vivien Kraus]

Hello,

Free software enables cooperation in a free society. More precisely, it
makes it easy for a user of a package to use a new version where the
personal information has been corrected. The thread in [1] questions
our handling of potential cases where a transgender contributor of Guix
or one of its packages requests to change their name. While it would be
nothing but cruel to deny such a request, I want to consider the
broader case of updating personal information in general.

If someone asks you to update your installation of a package to a new
tarball with updated personal information (or a new tag in a rewritten
history), then in a non-free society, you can only say, “Sorry, I’m not
allowed to”. In a free society, you’re allowed to, and you have tools
at your fingertips to make sure it’s harmless to you (diff with your
old version, if you are alone, or collectively check that it follows
semver, remember that it still has all the CVEs, and forget about the
old thing).

If accepting such a safe update makes a security system fire false
positives (such as, guix pull saying there’s a downgrade attack if
guix’ history has been safely rewritten), then it’s a limitation of the
security system. If it’s too much work to silence this warning for a
legitimate reason, then make an announcement about this particular
false positive and let the user proceed.

The guix users, I claim, would rather have a distribution of guix (and
the packages it provides) with accurate personal information, even if
it means to be annoyed for a moment with a security system.


Best regards, 

Vivien

[1] https://lists.gnu.org/archive/html/guix-devel/2024-03/msg00138.html

P.S. I am desensitized to eye-rolling when I talk about free
software ;)

Re: Guix role in a free society

[Tomas Volf]

[-- Attachment #1: Type: text/plain, Size: 849 bytes --]

On 2024-03-18 18:48:27 +0100, Vivien Kraus wrote:
> The guix users, I claim, would rather have a distribution of guix (and
> the packages it provides) with accurate personal information, even if
> it means to be annoyed for a moment with a security system.

Single data point: As a Guix user (and occasional contributor, albeit not a
committer), I would very much prefer a system that does not rewrite the history.
When someone wants to correct their name (for whatever reason), I would prefer
it to be done going forward, not retroactively.

I think making such broad statements without some empirical study is not great,
since it is, as you said yourself, just your claim not supported by anything (as
far as I can tell).

Tomas Volf

--
There are only two hard things in Computer Science:
cache invalidation, naming things and off-by-one errors.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: Guix role in a free society

[MSavoritias]

On 3/18/24 20:16, Tomas Volf wrote:
> On 2024-03-18 18:48:27 +0100, Vivien Kraus wrote:
>> The guix users, I claim, would rather have a distribution of guix (and
>> the packages it provides) with accurate personal information, even if
>> it means to be annoyed for a moment with a security system.
> Single data point: As a Guix user (and occasional contributor, albeit not a
> committer), I would very much prefer a system that does not rewrite the history.
> When someone wants to correct their name (for whatever reason), I would prefer
> it to be done going forward, not retroactively.
>
> I think making such broad statements without some empirical study is not great,
> since it is, as you said yourself, just your claim not supported by anything (as
> far as I can tell).
>
> Tomas Volf
>
> --
> There are only two hard things in Computer Science:
> cache invalidation, naming things and off-by-one errors.

It pretty easy to see who most people that use Guix agree with that 
actually. Check what the CoC says right here -> 
https://git.savannah.gnu.org/cgit/guix.git/tree/CODE-OF-CONDUCT

|We as members, contributors, and leaders pledge to make participation 
in our community a harassment-free experience for everyone, regardless 
of age, body size, visible or invisible disability, ethnicity, sex 
characteristics, gender identity and expression, level of experience, 
education, socio-economic status, nationality, personal appearance, 
race, caste, color, religion, or sexual identity and orientation. We 
pledge to act and interact in ways that contribute to an open, 
welcoming, diverse, inclusive, and healthy community. |

So since the Guix community have agreed to make it welcoming to 
everybody we have to take into account people that will want to change 
their names.

Social inclusion and people are above any tech ideals we may have.


We dont need to rewrite history at all also. There was a solution 
already by Gitlab which was also proposed in the other thread (for legal 
reasons) to do with UUIDs.


MSavoritias

Re: Guix role in a free society

[Tobias Alexandra Platen]

I am transgender and plan to contribute to Guix soon. Mostly submitting
packages and so on. I never saw any violation of the code of conduct,
but this does not mean they do not exist. 

On Mon, 2024-03-18 at 20:26 +0200, MSavoritias wrote:
> 
> On 3/18/24 20:16, Tomas Volf wrote:
> > On 2024-03-18 18:48:27 +0100, Vivien Kraus wrote:
> > > The guix users, I claim, would rather have a distribution of guix
> > > (and
> > > the packages it provides) with accurate personal information,
> > > even if
> > > it means to be annoyed for a moment with a security system.
> > Single data point: As a Guix user (and occasional contributor,
> > albeit not a
> > committer), I would very much prefer a system that does not rewrite
> > the history.
> > When someone wants to correct their name (for whatever reason), I
> > would prefer
> > it to be done going forward, not retroactively.
> > 
> > I think making such broad statements without some empirical study
> > is not great,
> > since it is, as you said yourself, just your claim not supported by
> > anything (as
> > far as I can tell).
> > 
> > Tomas Volf
> > 
> > --
> > There are only two hard things in Computer Science:
> > cache invalidation, naming things and off-by-one errors.
> 
> It pretty easy to see who most people that use Guix agree with that 
> actually. Check what the CoC says right here -> 
> https://git.savannah.gnu.org/cgit/guix.git/tree/CODE-OF-CONDUCT
> 
> > We as members, contributors, and leaders pledge to make
> > participation 
> in our community a harassment-free experience for everyone,
> regardless 
> of age, body size, visible or invisible disability, ethnicity, sex 
> characteristics, gender identity and expression, level of experience,
> education, socio-economic status, nationality, personal appearance, 
> race, caste, color, religion, or sexual identity and orientation. We 
> pledge to act and interact in ways that contribute to an open, 
> welcoming, diverse, inclusive, and healthy community. |
> 
> So since the Guix community have agreed to make it welcoming to 
> everybody we have to take into account people that will want to
> change 
> their names.
> 
> Social inclusion and people are above any tech ideals we may have.
> 
> 
> We dont need to rewrite history at all also. There was a solution 
> already by Gitlab which was also proposed in the other thread (for
> legal 
> reasons) to do with UUIDs.
> 
> 
> MSavoritias
> 
> 

Re: Guix role in a free society

[Richard Sent]

> It pretty easy to see who most people that use Guix agree with that
> actually. Check what the CoC says right here

I believe that Guix can continue to achieve a welcoming, harassment-free
environment even if we're not able to support repo authorship history
modification. (Or non-destructive attribution.)

I'm not in favor of (mandatory and global) UUIDs. To my understanding
there are two options for how they could be implemented:

a) UUIDs are used with .mailmap

  1) This doesn't solve the problem since .mailmap itself is also
  tracked in git. Any old names/aliases are still in the repo.

  2) This would mask the name change. To my knowledge unless someone is
  actively browsing .mailmap's log, the old name shouldn't appear. I
  understand why people may feel that's insufficient though.

  3) I don't believe any mechanism stops someone from choosing to
  do this already?

b) The UUID->Name mapping is stored out of band (GitLab's unimplemented
solution)

  1) This adds additional complication to development (need to fetch
  files over a network at some point, be sure you're using the right
  UUID even if you change machines, update your out of band copy
  regularly, etc).

We may be able to partially resolve b) but I doubt it's possible to turn
it into a "no-impact" process. It almost certainly would add steps for
new contributors. We don't want even more barriers to their first patch.

We could choose to allow people to opt-in to using UUIDs and also use
out-of-band storage, I suppose, but that would only help those who
already suspected they'd want to change their name, but didn't want to
change it at that moment. Otherwise a) would suffice.

Perhaps there are better options I'm not thinking of.

Would UUIDs be valid for the copyright notices at the top of files?

-- 
Take it easy,
Richard Sent
Making my computer weirder one commit at a time.

Re: Guix role in a free society

[Ludovic Courtès]

Hi MSavoritias,

MSavoritias <email@msavoritias.me> skribis:

> So since the Guix community have agreed to make it welcoming to
> everybody we have to take into account people that will want to change
> their names.

As I wrote earlier, several Guix contributors changed names in the past.
As a project, we always recognized the importance of calling people by
their chosen name and so those changes went smoothly.  So 💯 on your
comment above.

Now, I’d invite everyone to slow down on this conversation.  There are
important human and technical issues at stake, none of which is new or
specific to Guix or SWH.

Ludo’.

Re: Guix role in a free society

[Giovanni Biscuolo]

[-- Attachment #1: Type: text/plain, Size: 1479 bytes --]

Hello Vivien,

Vivien Kraus <vivien@planete-kraus.eu> writes:

> Free software enables cooperation in a free society. More precisely, it
> makes it easy for a user of a package to use a new version where the
> personal information has been corrected. The thread in [1] questions
> our handling of potential cases where a transgender contributor of Guix
> or one of its packages requests to change their name. While it would be
> nothing but cruel to deny such a request

Please do not frame the question that way because it's very different:
the original request is _not_ to use the correct personal information in
a new package to be distributed (and potentially used), the request is
to modify the _correct_ personal information (self) published in the
past by rewriting the git history of the SHW archived copy of the
software.

Guix contributors or package authors can change their personal
information - usually their name and email in copyright attribution(s)
and documentation - at any moment and that will be _authomatically_
propagated in all new Guix built artifacts and/or in the Guix git
repositories.

Also, git can _display_ a different name in git logs if instructed to to
so via .mailmap

The problem, let me call it a "rights clash", arises when pretenting
that "rewriting the past" is a right people can exercise, protected by
the european GDPR also.

[...]

Loving, Gio'

-- 
Giovanni Biscuolo

Xelera IT Infrastructures

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 849 bytes --]