Re: fvwm(1): replace GPL-licensed code, fix bugs and add improvements

Theo de Raadt wrote:
> Simon Wollwage <[email protected]> wrote:
> 
>> Stuart Henderson <[email protected]> writes:
>>
>>> On 2026/08/03 17:12, David Uhden Collado wrote:
>>>> Stuart Henderson wrote:
>>>>> On 2026/08/02 23:23, David Uhden Collado wrote:
>>>>>> I separated this work because the provenance and independence of the
>>>>>> replacement implementations require more careful review than ordinary
>>>>>> bug fixes.
>>>>>
>>>>> What _is_ the provenance of the replacement code implementations?
>>>>>
>>>> Robert Nation, Tom LaStrange, Charles Hines, Brady Montz, Dan Espen, Steve
>>>> Robbins, Paul Smith, Jason Tibbitts, Dominik Vogt, Bob Woodside, Maciej
>>>> Stachowiak, Greg J. Badros, and Andrew T. Veliath are free to pursue the
>>>> matter through the proper legal channels if they genuinely believe that I
>>>> have infringed their copyright. Since none of them appears to be based in
>>>> Spain, they would also have to establish jurisdiction and pursue the
>>>> corresponding international proceedings.
>>>>
>>>> A claimant would still need to prove ownership of the specific code at
>>>> issue, identify the allegedly copied expression, demonstrate actual
>>>> infringement rather than mere functional similarity, and establish
>>>> measurable damages caused by my non-commercial work. Unsupported accusations
>>>> and speculative threats are not evidence.
>>>>
>>>> I have no intention of being intimidated into abandoning independently
>>>> rewritten code merely because somebody invokes the GPL without demonstrating
>>>> what was supposedly copied, who owns it, or what legally actionable harm
>>>> occurred. If anyone believes they have a valid claim, they can present it
>>>> formally and prove it.
>>>
>>> Huh? Just asking where the code in your jumbo diff came from.
>>> I was hoping for something like "I wrote it myself" (although half
>>> expecting "some LLM wrote it").
>>>
>>> This very defensive rules-lawyer-ish statement does not fill me
>>> with confidence that this is code which OpenBSD can incorporate.
>>>
>>>
>>> Also there's your earlier "In order for FVWM to remain in the base
>>> system, a fork must be created and maintained" - well no. It could also
>>> remain how it is now...
>>
>> I am even more convinced now that this is LLM generated and should not
>> be merged because the provenance is even more fishy.
>>
> 
> I don't care if it is created by a LLM or not.  What bothers is who this
> is coming from.  There was a rant which completely fails to understand
> that the continued open-ness of our project depends upon known
> provenance and correct attribution of every item, and then completely
> turns the issue upside down.
> 
> I believe the cautious process going forward would be wariness,
> and never allow anything large into the tree from David Uhden Collado.
> 
> I will not spend one second more on this.
> 
Theo,

I understand that large patches require more caution and more review
time. What I find discouraging is that their size seems to create an
immediate presumption that the work is untrustworthy or LLM-generated.

I am trying to improve code that has received very little substantial
work for many years. Fixing security issues in the old FVWM code,
addressing accumulated bugs, and independently replacing the
GPL-licensed parts necessarily produces a substantial diff.

I documented the work as clearly as I could, including a plain-English
text file describing the changes and comments in both the patches and
the code. What the work still needs is someone with enough time to
review and test it.

I accept that a large patch is difficult to review. I do not think,
however, that its size alone justifies questioning its authorship or
provenance.

David.