• Freexian, security and Debian

    From Adrian Bunk@3:633/10 to All on Sunday, July 19, 2026 19:20:01
    -----BEGIN PGP SIGNED MESSAGE-----
    Hash: SHA512

    1. Overview
    - -----------

    This email covers two topics that are separate but have some intersection:

    There has to be a clearer separation between what Freexian activities
    are under oversight of Debian and can be called Debian and might happen
    on our infrastructure - and what is an external company competing for customers in our ecosystem.

    While there is no lack of DDs interested in earning money doing LTS and
    ELTS work, there is fewer interest in doing the same work unpaid in our security team instead. In recent weeks I got the impression that the two members who are doing nearly all the work in our security team seem to
    be struggling to keep up with the increased workload due to AI, and I am worried we might lose our security team to burnout.


    2. Debusine, and Freexian in general
    - ------------------------------------

    While it is laudable that Freexian has the explicit goal of helping
    Debian, Debian giving preferential treatment to one company might
    outweigh the benefits when this drives away other contributors.

    While there is valid criticism at the current use of GitLab in our core infrastructure,[1] using Debusine in our infrastructure would be worse:

    Similar to GitLab, Freexian SARL is a private for-profit company.

    Similar to Gitlab, Debusine is Open Source that is also commercially
    offered as SaaS.[2]

    Worse than GitLab Inc., which is a neutral player providing equal offers
    to everyone in the Debian ecosystem, Freexian SARL is competing directly
    with the businesses and employers of many other contributors (including
    many DDs) in our ecosystem.[3]

    There are technical reasons why Debusine is IMHO not suitable for our
    core infrastructure, but the point that this is a product of a company competing in our ecosystem alone should be sufficient that Debian keeps
    some distance from it.

    A company asking on debian-devel-announce for volunteers to work unpaid
    on their product is problematic.[4]

    Two of the accepted Google Summer of Code 2026 projects are improvements
    for Debusine.[5]

    Does Google know that they pay students to work on a company product,
    mentored by people working for that company?
    Are the Debian mentors paid by Freexian for the mentoring?

    Decisions which GSoC applications to select are made by the Outreach delegation.
    A majority of the Outreach delegates are receiving income from Freexian.[6,7] Did these delegates recuse themselves from GSoC due to the conflict of interest?


    3. ELTS
    - -------

    Freexian ELTS is a commercial offer by an external company that has to
    be clearly separate from anything called Debian.

    It is problematic that information about Freexian ELTS is offered on the
    same company website as the information about Debian LTS,[8] and that
    Debian gives the impression of official status of ELTS by listing it in distro-info and distro-info-data.

    It should be made clearer to our users what is Debian and what is a
    commercial offer by an external company, and Debian LTS information
    should be moved to our website.


    4. LTS
    - ------

    After an intervention from Freexian,[9,10] the regular support of
    bookworm was shortened from what was originally planned - while longer
    full support would have been better for our users, it would have been
    worse for Freexian.

    Until recently Freexian was transparent about the monthly Debian LTS
    payments to individual contributors (multiply with ? 85/hour[6]),[11]
    but Freexian now removed this transparency.[12]

    When non-profits like the Linux Foundation and companies sponsor
    something that is called "Debian Long Term Support"[14] and is provided
    on our (sponsored) infrastructure, we need more transparency and not
    less.

    It would be appropriate that Freexian sends detailed monthly financial
    reports to a public Debian mailing list, including stating for each work
    item invoiced by a Freexian contributor how much Debian LTS money was
    spent on it.

    Sending such a financial report for June 2026 at the same time as the
    report about work done in June 2026 would be a good start.

    With approximately ? 300k annually[13] Debian LTS is among the larger
    budgets in Debian.

    Our publicity refers to "Debian and its LTS Team", and Debian does
    encourage funding Debian LTS.[14] Debian recommending to fund Debian LTS instead of recommending donating to Debian might be redirecting some
    donations from Debian to Debian LTS.

    Since the topic that Freexian no longer wanted my services will
    invariably come up:

    While there are many people who are doing good work in LTS (and ELTS)
    and where the hours invoiced are in a reasonable relation to the work
    done, a not so small part of money goes to the half dozen Freexian contributors where the amount of hours invoiced always felt like an
    order of magnitude larger than the quantity and quality of work
    delivered. My conflict with Freexian management was around me trying
    to get Freexian to do something about it.

    Some examples:

    Freexian contributors who might suddenly show up on the last day of a
    month, do some rather trivial work, and invoice 30 or 50 hours every month.[15]

    A Freexian contributor invoices 19.5 hours[16] (? 1657.50) from Debian
    LTS (plus additional hours from ELTS) with a monthly report[17] that is largely made up:

    The zfs-linux DLA was issued, but this was the 4th month where money was invoiced for working on this package[17,18,19,20] and no other DLA was
    issued by this contributor during that time.

    "Worked on phpmyadmin, zabbix, and atril but couldn?t complete them.
    Will get back to it next month." There is not much trace of any work
    done (or even a DLA issued) on these packages by this contributor.

    Most front desk work (mainly triaging) is done with git commits, but
    there is not a single commit by this contributor in the git tree[21]
    during this week. The list of what was claimed to be triaged and Auto
    EOL?d also looks quite similar to what was claimed to be worked on
    (and invoiced) a month earlier.[18]

    The cheeky "PS: it?s not. :)" for claimed debugging work regarding the
    xz backdoor?
    This would also have created git commits.
    LTS was marked not affected by a member of our security team,[22]
    and the ELTS releases were marked not affected by a different Freexian contributor 2 weeks later.[23]

    This monthly report was an example not an exception for this Freexian contributor, and a point could be made that a DD who is defrauding
    Debian LTS is defrauding Debian and should therefore be expelled
    from Debian.

    After sending analysis of more than one monthly report to the Freexian managers responsible for LTS and ELTS I gave up, and the person is still
    a Freexian contributor today.

    A Freexian contributor, who as Debian maintainer of the package in
    question should already know it quite well, invoices 130.75 hours[24,25,26,27] (? 11 113.75) from Debian LTS (plus additional hours from ELTS) without
    ever delivering an LTS update for the package in LTS (buster) and
    without doing any other significant work during these months.[28,29,30,31]

    The response I got from the Freexian managers responsible for LTS and ELTS:

    Santiago and I have communicated with <contributor> and we are satisfied
    that the work reported is appropriate to the hours invoiced for July,
    and the same going back to April. Based on that, we see no need for
    further action.


    5. Workload of our Security Team (and LTS)
    - ------------------------------------------

    For all incoming CVEs, our security team does:
    - - map it to the Debian package name (or mark as not-for-us)
    - For a CVE in a Debian package our security team does:
    - link to the upstream advisory
    - often search for the upstream commit(s) fixing it,
    and link to them in the security tracker
    - triage whether the CVE warrants a DSA for the package

    All CVE triaging and tracking, and work assignment for both our security
    team and LTS, happens in the same git tree. It is worth noticing that
    the two members who are doing most work in our security team together
    average nearly 50 commits per day, mostly with CVE triaging and
    tracking.[32]

    Initial LTS triaging is not hard based in this work already done by our security team.

    The number of DSAs went from ~20 per month a year ago to ~60 per month today. That's an average of 2 DSAs our security team issues every day.

    Different from what some people anticipated in discussions before LTS:
    Actual work in LTS is often just taking what the security team has
    already published as a DSA one release further back.
    Sometimes backporting is some work, but this is not done in the hard
    cases - when something is hard usually either the package is listed as
    no longer supported or the CVE is marked as "ignored" with a comment
    like "too invasive to fix".
    This hard work would be the part where providing 5 years of support
    would not work unpaid, due to both the skills and the effort required.

    To be fair, people with the qualifications to fix hard issues (including
    doing hard backports) in security are not available at the hourly rate
    paid in LTS, and the security teams of commercial distributions
    providing long-term support are also far better resourced.

    Debian LTS pays security team work done by the security team member who
    is also a Freexian contributor.[33,34,35,36]
    Why does Debian LTS pay one security team member, but not the ones who
    are doing most of the work?

    At the hourly rate paid in LTS the money available equals two full-time positions, and one option would be to offer the members of our security
    team who are doing nearly all of the work this money so that they can
    quit (or at least reduce the hours at) their current work and work
    full-time on Debian security including LTS.

    Other solutions are possible, but Debian continuing to offer the same
    work paid and unpaid in different teams is not sustainable.



    [1] but there was no better option available at the time
    [2] https://www.freexian.com/services/debusine/
    [3] https://www.freexian.com/services/
    [4] https://lists.debian.org/debian-devel-announce/2025/08/msg00005.html
    [5] https://summerofcode.withgoogle.com/programs/2026/organizations/debian
    [6] https://www.freexian.com/lts/debian/details/#faq
    [7] https://www.freexian.com/about/team/
    [8] https://www.freexian.com/lts/
    [9] https://lists.debian.org/debian-release/2025/09/msg00507.html
    [10] https://lists.debian.org/debian-release/2025/09/msg00509.html
    [11] https://www.freexian.com/blog/debian-lts-report-2025-09/
    [12] https://www.freexian.com/blog/debian-lts-report-2026-05/
    [13] https://www.freexian.com/lts/debian/
    [14] https://lists.debian.org/debian-announce/2026/msg00007.html
    [15] LTS and ELTS combined
    [16] https://www.freexian.com/blog/debian-lts-report-2024-03/
    [17] https://web.archive.org/web/20240620143011/https://utkarsh2102.org/posts/foss-in-march-24/
    [18] https://web.archive.org/web/20240620141659/https://utkarsh2102.org/posts/foss-in-feb-24/
    [19] https://web.archive.org/web/20260414090616/https://utkarsh2102.org/posts/foss-in-jan-24/
    [20] https://web.archive.org/web/20260213185556/https://utkarsh2102.org/posts/foss-in-dec-23/
    [21] https://salsa.debian.org/freexian-team/extended-lts/security-tracker
    [22] https://salsa.debian.org/freexian-team/extended-lts/security-tracker/-/commit/49a400ede5d
    [23] https://salsa.debian.org/freexian-team/extended-lts/security-tracker/-/commit/c3f49879ae1
    [24] https://www.freexian.com/blog/debian-lts-report-2024-04/
    [25] https://www.freexian.com/blog/debian-lts-report-2024-05/
    [26] https://www.freexian.com/blog/debian-lts-report-2024-06/
    [27] https://www.freexian.com/blog/debian-lts-report-2024-07/
    [28] https://lists.debian.org/debian-lts/2024/05/msg00004.html
    [29] https://lists.debian.org/debian-lts/2024/06/msg00002.html
    [30] https://lists.debian.org/debian-lts/2024/07/msg00014.html
    [31] https://lists.debian.org/debian-lts/2024/08/msg00002.html
    [32] https://salsa.debian.org/security-tracker-team/security-tracker/-/commits/master
    [33] https://www.freexian.com/blog/debian-lts-report-2026-01/
    [34] https://dl.gambaru.de/blog/202601_LTS_ELTS_report.txt
    [35] https://www.freexian.com/blog/debian-lts-report-2026-03/
    [36] https://dl.gambaru.de/blog/202603_LTS_ELTS_report.txt
    -----BEGIN PGP SIGNATURE-----

    iQIzBAEBCgAdFiEEOvp1f6xuoR0v9F3wiNJCh6LYmLEFAmpdAe0ACgkQiNJCh6LY mLGt4g/9GCf4QNGYtBwlv2AHuuiCCDv78+9st4w5XwYr7qH6Llta3eH3Ppw4xHrz yZP5sprKPjFgLl7kC06pePK5nyyhBof/7EopIlhaEg0DdqIje6tEc35FvfBIyaQ6 5P8qq/N25lW7WDea6u/53O2d5o1YPTZZnmyrTwSXJJcHBcydHBOZ+T/sXU+zBPSU 4iV3LUDSMNwEuivbPc33pxh7rev4eKn195EGtuY9Do/2sD1wTrI0e3nN8T84xMd9 ZCsGvy7uGmLugxuwyzYXrZVdcD5dNRrk4j43XMrkAMxcwczSXxxw/RDhjp+q+d6Y YdL7tXw/yuOTzBu/sIvk2GSof/ctq7QYWxd+l5/zVQcGkfGpLUNDVKNmpAak4uze U4woofBQxFJSSrZrnGfaS1+mgPnFUD3HNRIDzMfLrWuFUEua+tMiUoSRrx0bXXxn h37Xh3fvRWcb0mo5kKSfJTdRJ3ejKaOzs1zGYKZ2eSa1aojFgK/0vfEvMMsqgu0b F+zYeA/lRPzNEv3m53XcKQbyICb8agcuMVhGgiDnrIYCUaEjAE5DTXq03Z+hHIJz nt90LHlvZ63YJbFWIaq33hZHIFKc2YGlAZdfD7r/RUTO+bqok3ioRYOOeQqvdrql lFRbnGtdlS0nVgAbZHM19ftB9x7tCU11AxkGKCSRt3jFdLaZy08=
    =ehCS
    -----END PGP SIGNATURE-----

    --- PyGate Linux v1.5.18
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Joerg Jaspert@3:633/10 to All on Tuesday, July 21, 2026 00:30:02
    On 18025 March 1977, Adrian Bunk wrote:

    I do agree that something like Freexian and it's effects (both, good and
    bad) do warrant some project internal discussions. The size (small) and workload (huge) of our security team (and other core teams) is known.
    How that could be improved (or not) with the kind of fundraising done is
    a topic that had strong positions in the past. And for sure will still
    have that now.


    I do not think your mail does help this case. It does read quite a lot
    of "they kicked me, now I fight them". That is a bad start.

    But the most interesting part in my first read of it is this:

    This monthly report was an example not an exception for this Freexian contributor, and a point could be made that a DD who is defrauding
    Debian LTS is defrauding Debian and should therefore be expelled
    from Debian.

    First, we do have a process around somewhere, if one DD thinks another
    should be out of the project. Not used often, thankfully, but existing.

    Second, while the above is written without names, it shouldn't be hard
    for people involved to find out who is meant. Something like slander, defamation or similar could get lawyers involved. Doesn't sound like a
    way to take. Please be careful what you write on our public lists.

    --
    bye, Joerg

    --- PyGate Linux v1.5.18
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Hector Oron@3:633/10 to All on Tuesday, July 21, 2026 13:50:01
    Hello Project,

    I just wanted to share a fund that could be requested and if you think
    Debian should apply for it: https://www.isocfoundation.org/grant-programme/common-good-cyber-fund/

    How could Debian use such a fund (since all of it would require
    financial control and lots of paperwork)?

    Regards
    --
    H‚ctor Or˘n -.. . -... .. .- -. -.. . ...- . .-.. --- .--. .
    .-.

    --- PyGate Linux v1.5.18
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Adrian Bunk@3:633/10 to All on Tuesday, July 21, 2026 18:10:01
    On Tue, Jul 21, 2026 at 12:13:11AM +0200, Joerg Jaspert wrote:
    On 18025 March 1977, Adrian Bunk wrote:

    I do agree that something like Freexian and it's effects (both, good and
    bad) do warrant some project internal discussions. The size (small) and workload (huge) of our security team (and other core teams) is known. How that could be improved (or not) with the kind of fundraising done is a topic that had strong positions in the past. And for sure will still have that
    now.


    I do not think your mail does help this case.

    What else should I have written in your opinion to help this case?

    My trigger was the "I am worried we might lose our security team to
    burnout" part around the latest point releases, which removed "nothing"
    from the options.

    It does read quite a lot of
    "they kicked me, now I fight them". That is a bad start.

    I have been thinking for weeks what and how to write.

    If I omit this part, they might slander me behind my back with the same
    "they kicked me, now I fight them" - and I won't have an opportunity to
    respond to that.

    My conclusion was that writing everything in the initial email was my
    only option.

    But the most interesting part in my first read of it is this:

    This monthly report was an example not an exception for this Freexian contributor, and a point could be made that a DD who is defrauding
    Debian LTS is defrauding Debian and should therefore be expelled
    from Debian.

    First, we do have a process around somewhere, if one DD thinks another
    should be out of the project. Not used often, thankfully, but existing.

    ? 4.1.3. is the process I would have used if that was my intention.

    I am a strong proponent of a fair trial.
    There is precedent for DAM expelling a member without prior warning and without hearing the accused before expulsion.

    In this case DAM would anyway not have been an option, since DAM cannot
    decide whether Debian LTS money is Debian money - which would be the
    most relevant question for a judgement.

    This problem is also primarily management inaction and I am not
    particularly interested in going after individual Freexian contributors
    in Debian, but vague hints instead of providing clear examples would
    lead nowhere.

    Second, while the above is written without names, it shouldn't be hard for people involved to find out who is meant. Something like slander, defamation or similar could get lawyers involved. Doesn't sound like a way to take. Please be careful what you write on our public lists.

    There is no good option when bringing up this kind of problems,
    and it is possible that I failed to find the least bad option.

    I tried to stay as close as possible at what is supported by public information I linked to, and provided only as many examples as I
    considered necessary for demonstrating problems.

    Is this our LTS Team as our publicity says, or an external company doing business on our infrastructure without being accountable to Debian that
    might get lawyers involved?

    For me Debian LTS was always also a part of Debian, not work for a
    normal company where one just has to ignore everything bad going on
    inside the company.

    bye, Joerg

    cu
    Adrian

    --- PyGate Linux v1.5.18
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Adrian Bunk@3:633/10 to All on Wednesday, July 22, 2026 13:30:01
    On Tue, Jul 21, 2026 at 12:24:25PM +0200, Hector Oron wrote:
    ...
    https://www.isocfoundation.org/grant-programme/common-good-cyber-fund/

    How could Debian use such a fund

    The first problem would be the requirement "that the applicant is a
    qualifying nonprofit organization", which is the old issue that Debian
    is not a legal entity.

    Perhaps SPI might be sufficient.

    (since all of it would require
    financial control and lots of paperwork)?

    This is the longstanding problem that a non-profit project of our size
    and asking for donations should really have a proper budget process and
    annual financial reports on our website.

    And then something resembling a fundraising strategy.
    Even the basic stuff like a "Donate" link on our website is missing.

    What do we want?
    What do we not want?
    What do we need for what we want?

    ...
    and if you think Debian should apply for it
    ...

    It tends to produce poor (sometimes even negative) outcomes when
    a project starts with having (or applying for) money.

    The program you are linking to would be one person paid full-time for
    two years, and then there would have to be both a suitable project
    planned, and confidence that paying money there won't have a negative
    impact on the unpaid volunteers working in that area.

    A project that might fit the last bullet point of Objective 1 in this
    grant would be to make our own infrastructure more resilient, two years wouldn't be too much for designing, discussing and implementing improvements.

    But such a project would have to be properly planned before applying.
    And it would need someone available with the required skills in
    security and AI.

    Regards

    cu
    Adrian

    --- PyGate Linux v1.5.18
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Pierre-Elliott BĂ©cue@3:633/10 to All on Wednesday, July 22, 2026 16:40:01
    Just jumping on this bit.
    Adrian Bunk <bunk@debian.org> wrote on 19/07/2026 at 18:57:21+0200:
    Worse than GitLab Inc., which is a neutral player providing equal
    offers to everyone in the Debian ecosystem, Freexian SARL is competing directly with the businesses and employers of many other contributors (including many DDs) in our ecosystem.[3]
    This does neither seem true (about GitLab being neutral) nor really
    relevant.
    Many of us have employers that are competing with each others.
    --
    PEB


    --- PyGate Linux v1.5.18
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Lucas Nussbaum@3:633/10 to All on Wednesday, July 22, 2026 16:40:01
    Hi,

    On 19/07/26 at 19:57 +0300, Adrian Bunk wrote:
    2. Debusine, and Freexian in general
    ------------------------------------

    While it is laudable that Freexian has the explicit goal of helping
    Debian, Debian giving preferential treatment to one company might
    outweigh the benefits when this drives away other contributors.

    A long time ago, I was the DPL when Debian LTS was set up.

    I was very careful at the time to defend a setup where mutiple actors
    could collaborate on Debian LTS. I think that this still holds: if
    someone if sufficiently unhappy with Freexian's contributions to LTS,
    there's space to setup other actors, with their own funding models.

    While there is valid criticism at the current use of GitLab in our core infrastructure,[1] using Debusine in our infrastructure would be worse:

    Similar to GitLab, Freexian SARL is a private for-profit company.

    Similar to Gitlab, Debusine is Open Source that is also commercially
    offered as SaaS.[2]

    Worse than GitLab Inc., which is a neutral player providing equal offers
    to everyone in the Debian ecosystem, Freexian SARL is competing directly with the businesses and employers of many other contributors (including
    many DDs) in our ecosystem.[3]

    Unlike GitLab, Debusine does not have an Open Core business model, where
    the editor could decide to remove important features from the Community
    Edition we are using. It is pure GPLv3 with no CLA. I think that's a key distinction.

    Lucas

    --- PyGate Linux v1.5.18
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Hector Oron@3:633/10 to All on Wednesday, July 22, 2026 22:00:01
    Hello Adrian,

    El mi‚, 22 jul 2026 a las 13:21, Adrian Bunk (<bunk@debian.org>) escri
    bi˘:

    The first problem would be the requirement "that the applicant is a qualifying nonprofit organization", which is the old issue that Debian
    is not a legal entity.

    Perhaps SPI might be sufficient.

    Yes, SPI qualifies.

    (since all of it would require
    financial control and lots of paperwork)?

    This is the longstanding problem that a non-profit project of our size
    and asking for donations should really have a proper budget process and annual financial reports on our website.

    That should be covered by SPI, the non-profit, Debian is associated project
    .

    The program you are linking to would be one person paid full-time for
    two years, and then there would have to be both a suitable project
    planned, and confidence that paying money there won't have a negative
    impact on the unpaid volunteers working in that area.

    A project that might fit the last bullet point of Objective 1 in this
    grant would be to make our own infrastructure more resilient, two years wouldn't be too much for designing, discussing and implementing improveme
    nts.

    But such a project would have to be properly planned before applying.
    And it would need someone available with the required skills in
    security and AI.

    When reading your email about "security team burnout", I thought this
    fund could be an option to mitigate such concern, however, how could
    make it fit and work for the project? Would it be valuable?

    Regards
    --
    H‚ctor Or˘n -.. . -... .. .- -. -.. . ...- . .-.. --- .--. .
    .-.

    --- PyGate Linux v1.5.18
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Simon Josefsson@3:633/10 to All on Wednesday, July 22, 2026 22:00:01
    Lucas Nussbaum <lucas@debian.org> writes:
    Hi,

    On 19/07/26 at 19:57 +0300, Adrian Bunk wrote:
    2. Debusine, and Freexian in general
    ------------------------------------

    While it is laudable that Freexian has the explicit goal of helping
    Debian, Debian giving preferential treatment to one company might
    outweigh the benefits when this drives away other contributors.

    A long time ago, I was the DPL when Debian LTS was set up.

    I was very careful at the time to defend a setup where mutiple actors
    could collaborate on Debian LTS. I think that this still holds: if
    someone if sufficiently unhappy with Freexian's contributions to LTS,
    there's space to setup other actors, with their own funding models.
    Would that even be possible if Debian buildds were to use Debusine?
    Doing so seems like a corporate-takeover of an essential Debian-internal service.
    I guess the close association between Debian-internal matters (e.g., LTS
    or buildd) with one external company is what may annoy people. It calls
    for transparency.
    Don't get me wrong: in this particular case, I'm for replacing our aging
    buildd infrastructure with Debusine which is well documented and appears
    on the surface to be better engineered. I don't have anything per se
    against corporate takeovers, and find that they often lead to more
    activity and progress, which is good. But we should be aware of what we
    are doing.
    /Simon


    --- PyGate Linux v1.5.18
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Adrian Bunk@3:633/10 to All on Wednesday, July 22, 2026 22:20:01
    On Wed, Jul 22, 2026 at 06:42:31PM +0200, Hector Oron wrote:
    Hello Adrian,

    El mi‚, 22 jul 2026 a las 13:21, Adrian Bunk (<bunk@debian.org>) escribi˘:

    The first problem would be the requirement "that the applicant is a qualifying nonprofit organization", which is the old issue that Debian
    is not a legal entity.

    Perhaps SPI might be sufficient.

    Yes, SPI qualifies.

    (since all of it would require
    financial control and lots of paperwork)?

    This is the longstanding problem that a non-profit project of our size
    and asking for donations should really have a proper budget process and annual financial reports on our website.

    That should be covered by SPI, the non-profit, Debian is associated project.

    What about money in other TOs?

    Is "A current operational budget" something Debian has?

    The program you are linking to would be one person paid full-time for
    two years, and then there would have to be both a suitable project
    planned, and confidence that paying money there won't have a negative impact on the unpaid volunteers working in that area.

    A project that might fit the last bullet point of Objective 1 in this
    grant would be to make our own infrastructure more resilient, two years wouldn't be too much for designing, discussing and implementing improvements.

    But such a project would have to be properly planned before applying.
    And it would need someone available with the required skills in
    security and AI.

    When reading your email about "security team burnout", I thought this
    fund could be an option to mitigate such concern, however, how could
    make it fit and work for the project? Would it be valuable?

    The general problem with such grants is that they are for additional features[1] - this is temporary funding of additional work, not
    long-term funding of everyday work.

    Regards

    cu
    Adrian

    [1] with a very loose definition of "features",
    e.g. new hardware might get funded

    --- PyGate Linux v1.5.18
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Adrian Bunk@3:633/10 to All on Thursday, July 23, 2026 15:20:01
    On Wed, Jul 22, 2026 at 04:05:32PM +0200, Pierre-Elliott B‚cue wrote:
    Just jumping on this bit.

    Adrian Bunk <bunk@debian.org> wrote on 19/07/2026 at 18:57:21+0200:

    Worse than GitLab Inc., which is a neutral player providing equal
    offers to everyone in the Debian ecosystem, Freexian SARL is competing directly with the businesses and employers of many other contributors (including many DDs) in our ecosystem.[3]

    This does neither seem true (about GitLab being neutral) nor really
    relevant.

    The Debian services Frexian offers[1] are in direct competition to the
    Debian services offered by many Debian contributors or their employers.

    And if Debusine would become the official way for building Debian, then
    the SaaS offer of the company maintaining Debusine would be a unique
    selling point.

    GitLab is in a different market.

    Many of us have employers that are competing with each others.

    Other competitors did not receive ? 485k taxpayer money to develop their product by presenting it as "aimed at modernizing Debian's infrastructure"
    so that Debian can "attract [and] mantain its community of highly
    skilled volunteers".[2]

    cu
    Adrian

    [1] https://www.freexian.com/services/
    [2] https://ted.europa.eu/en/notice/-/detail/524641-2023

    --- PyGate Linux v1.5.18
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Sean Whitton@3:633/10 to All on Thursday, July 23, 2026 15:40:01
    Joerg Jaspert [21/Jul 12:13am +02] wrote:
    But the most interesting part in my first read of it is this:

    This monthly report was an example not an exception for this Freexian
    contributor, and a point could be made that a DD who is defrauding
    Debian LTS is defrauding Debian and should therefore be expelled
    from Debian.

    First, we do have a process around somewhere, if one DD thinks another
    should be out of the project. Not used often, thankfully, but existing.

    Second, while the above is written without names, it shouldn't be hard
    for people involved to find out who is meant. Something like slander, defamation or similar could get lawyers involved. Doesn't sound like a
    way to take. Please be careful what you write on our public lists.
    Right. Adrian, it was irresponsible of you to post this to a public
    list. If you thought it was very important to include it (and I'm not convinced it was), you could have sent it as a reply but addressed only
    to debian-private@.
    --
    Sean Whitton


    --- PyGate Linux v1.5.18
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Pierre-Elliott BĂ©cue@3:633/10 to All on Friday, July 24, 2026 01:00:01
    Adrian Bunk <bunk@debian.org> wrote on 23/07/2026 at 11:01:58+0200:
    On Wed, Jul 22, 2026 at 04:05:32PM +0200, Pierre-Elliott B‚cue wrote:
    Just jumping on this bit.

    Adrian Bunk <bunk@debian.org> wrote on 19/07/2026 at 18:57:21+0200:

    Worse than GitLab Inc., which is a neutral player providing equal
    offers to everyone in the Debian ecosystem, Freexian SARL is competing
    directly with the businesses and employers of many other contributors
    (including many DDs) in our ecosystem.[3]

    This does neither seem true (about GitLab being neutral) nor really
    relevant.

    The Debian services Frexian offers[1] are in direct competition to the
    Debian services offered by many Debian contributors or their
    employers.
    And all our employers are all in competition with each other already?
    This makes no sense.
    And if Debusine would become the official way for building Debian, then
    the SaaS offer of the company maintaining Debusine would be a unique
    selling point.
    Debusine can be installed on premise so this is blatantly false.
    GitLab is in a different market.
    This makes no sense and still is not relevant.
    Many of us have employers that are competing with each others.

    Other competitors did not receive ? 485k taxpayer money to develop
    their product by presenting it as "aimed at modernizing Debian's infrastructure" so that Debian can "attract [and] mantain its
    community of highly skilled volunteers".[2]
    Other competitors could have asked for funding for their project
    provided it adhered to the conditions described here[1] or there[2].
    I really don't know what you are trying to prove here, but you are doing
    it badly, in my opinion.
    To be fair, I agree with Steve, you're obviously stuck into a
    finger-pointing name and shame loop against some people for whatever
    reason of yours you have. This is not sane and this is not really
    acceptable either.
    I'll stick to the sane aspect though, I'd recommend that you take a step
    back, breath a good bit of fresh air.
    Bests,
    --
    PEB
    [1] https://interoperable-europe.ec.europa.eu/collection/open-source-observatory-osor/document/funding-open-source-case-study-sovereign-tech-fund
    [2] https://www.sovereign.tech/programs/fund


    --- PyGate Linux v1.5.18
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Hector Oron@3:633/10 to All on Saturday, July 25, 2026 15:10:01
    Hello,
    El mi‚, 22 jul 2026, 22:14, Adrian Bunk <bunk@debian.org> escribi˘:
    This is the longstanding problem that a non-profit project of our size
    and asking for donations should really have a proper budget process and annual financial reports on our website.

    That should be covered by SPI, the non-profit, Debian is associated
    project.

    What about money in other TOs?

    Every TO should have a yearly financial report, I assume they need to keep legal compliance.
    Is "A current operational budget" something Debian has?

    No, a community usually is not run by budgets.
    When reading your email about "security team burnout", I thought this
    fund could be an option to mitigate such concern, however, how could
    make it fit and work for the project? Would it be valuable?

    The general problem with such grants is that they are for additional features[1] - this is temporary funding of additional work, not
    long-term funding of everyday work.

    Thanks for your thoughts on this, I agree it is a complex fit.
    Regards


    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Adrian Bunk@3:633/10 to All on Saturday, July 25, 2026 15:10:01
    On Sat, Jul 25, 2026 at 10:34:39AM +0200, Hector Oron wrote:
    Hello,

    El mi‚, 22 jul 2026, 22:14, Adrian Bunk <bunk@debian.org> escribi˘:

    This is the longstanding problem that a non-profit project of our size
    and asking for donations should really have a proper budget process and annual financial reports on our website.

    That should be covered by SPI, the non-profit, Debian is associated
    project.

    What about money in other TOs?


    Every TO should have a yearly financial report, I assume they need to keep legal compliance.

    Is "A current operational budget" something Debian has?


    No, a community usually is not run by budgets.
    ...

    As soon as money is involved, there is usually a budget for the
    financial part.

    At least I am used to even small (e.g. sports) clubs with a dozen
    members making a budget balancing income and expenses for the next year,
    which is the basis for decisions like whether to change the membership fee.

    Regards

    cu
    Adrian

    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Adrian Bunk@3:633/10 to All on Wednesday, July 29, 2026 19:00:01
    On Wed, Jul 22, 2026 at 03:20:42PM +0200, Lucas Nussbaum wrote:
    Hi,

    Hi Lucas,

    On 19/07/26 at 19:57 +0300, Adrian Bunk wrote:
    2. Debusine, and Freexian in general
    ------------------------------------

    While it is laudable that Freexian has the explicit goal of helping Debian, Debian giving preferential treatment to one company might
    outweigh the benefits when this drives away other contributors.

    A long time ago, I was the DPL when Debian LTS was set up.

    I was very careful at the time to defend a setup where mutiple actors
    could collaborate on Debian LTS. I think that this still holds: if
    someone if sufficiently unhappy with Freexian's contributions to LTS,
    there's space to setup other actors, with their own funding models.

    it is not wrong that there is in practice exactly one company/delegation/... for something like Debian LTS, many companies and individuals competing
    for Debian LTS sponsors wouldn't be better.

    Currently the top news on our homepage talks about "Debian and its LTS
    Team" and suggests to our users to consider funding it, and "Freexian is
    just an intermediary company" on the page in our wiki this news links to
    does not sound as if this sponsorship we endorse is money Freexian could
    use for anything other than LTS.

    My view of what is actually happening is that Freexian is developing
    from an unofficial for-profit subsidary of Debian with some transparency
    to a normal closed external company.

    While there is valid criticism at the current use of GitLab in our core infrastructure,[1] using Debusine in our infrastructure would be worse:

    Similar to GitLab, Freexian SARL is a private for-profit company.

    Similar to Gitlab, Debusine is Open Source that is also commercially offered as SaaS.[2]

    Worse than GitLab Inc., which is a neutral player providing equal offers to everyone in the Debian ecosystem, Freexian SARL is competing directly with the businesses and employers of many other contributors (including many DDs) in our ecosystem.[3]

    Unlike GitLab, Debusine does not have an Open Core business model, where
    the editor could decide to remove important features from the Community Edition we are using. It is pure GPLv3 with no CLA. I think that's a key distinction.

    GPLv3 would not make a difference if the company decided that only the Community Edition would be available self-hosted, with the full version
    only offered as SaaS.

    An obvious example would be Debian adopting a strong anti-AI position,
    since there is likely a good business case for AI support in the SaaS offer.

    It is also not rare (e.g. Atlassian, SourceForge) that the self-hosted
    version is discontinued. Then we are in the alioth situation in any case.

    Lucas

    cu
    Adrian

    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Adrian Bunk@3:633/10 to All on Wednesday, July 29, 2026 19:00:01
    On Thu, Jul 23, 2026 at 02:15:15PM +0100, Sean Whitton wrote:
    ...
    Right. Adrian, it was irresponsible of you to post this to a public
    list. If you thought it was very important to include it (and I'm not convinced it was), you could have sent it as a reply but addressed only
    to debian-private@.

    Apologies for that, I realize that this would have been a better option.

    Sean Whitton

    cu
    Adrian

    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Adrian Bunk@3:633/10 to All on Wednesday, July 29, 2026 19:00:01
    On Fri, Jul 24, 2026 at 12:17:31AM +0200, Pierre-Elliott B‚cue wrote:
    ...
    Other competitors could have asked for funding for their project
    provided it adhered to the conditions described here[1] or there[2].

    I really don't know what you are trying to prove here, but you are doing
    it badly, in my opinion.
    ...

    My understanding of your point is that even if Debian adopts a strong
    anti-AI position, then it would still be fine for you if a company would
    get half a million Euro granted by calling a project "Debian AI".

    My point is that organizations usually protect their trademarks and
    reputation by maintaining control over what happens in their name.

    If you start an open source project that includes "git" in its name,
    the lawyers of the Software Freedom Conservancy might contact you and
    explain the git trademark to you.

    Several years ago a non-profit external package repository run by a DD
    changed its name after an intervention by the DPL.


    My larger point is that much about money in Debian and in the name of
    Debian is intransparent and therefore inefficient.

    While searching for the number when replying to your previous email I
    realized that when counting both projects for Debian and projects that
    use use our distribution for an example implementation, [2] has granted
    over ? 1.5 million to Debian-related projects during the past 36 months.

    Why aren't we having a project-wide discussion in Debian where we need
    and want money, including talking with the people at [2] how much input
    from Debian they accept or even appreciate for their decisions?

    Do we want money for projects or are there positions where we might want
    to have people receive a monthly salary through their fellowship or both
    or none?

    Our own finances would also benefit from annual reports, having an
    annual budget, and some plan how to do more fundraising if we want
    to spend more money than the expected income.


    Bests,
    ...
    [2] https://www.sovereign.tech/programs/fund

    cu
    Adrian

    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)