[Development] QUIP-18 and file "reclassification"

[Volker Hilsheimer]

> On 19 Dec 2024, at 21:31, Jörg Bornemann via Development <development at qt-project.org> wrote:
> 
> On 12/19/24 12:06 PM, Marc Mutz via Development wrote:
> 
>> 1/
>> I today refused to perform a re-licensing of a .cpp file being renamed
>> to .qdoc.¹ QUIP-18² has no provisions for files transitioning between
>> what QUIP-18 calls file "classifications"³, even if they have different
>> required license specifiers.
>> Can someone please clarify the process here?
> 
> Inofficial stance, IANAL disclaimer applies: the file in question has a "Copyright (C) 2016 The Qt Company Ltd." note. The copyright and IP holder has decided to change the license from one liberal license to another. The practical effect this has on anyone is zero. But it makes the license checker happy, which sees ".qdoc file extension, this must be documentation".
> 
> Even without the sole TQtC copyright the CLA has been signed by all contributors and to my understanding enables us to change the license. There's no official process to my knowledge. But some official clarification might be in order here.
> 
>> 2/
>> In the same vein, it seems to me to be illogical to require to have a
>> class-only .qdoc file have a different license from the equivalent .cpp
>> file. These files may change from .cpp to .qdoc back back several times,
>> depending on whether there is some C++ code in the file or not (doesn't
>> have to be the implementation of a member function, could just be a
>> static_assert() that we don't want to waste CPU cycles on by including
>> it in the header).
>> Should be make a distinction between "pure" .qdoc files and those that
>> are essentially .cpp files that just lack (C++) content?
> 
> It seems logical to me to categorize .cpp files as source code and .qdoc as documentation. If the files *really* tend to oscillate between .cpp and .qdoc, let the license oscillate as well. Proposed process: match what has been agreed on in QUIP-18.
> 
> I'd refrain from adding heuristics to the license checker to determine whether a .qdoc file looks a bit too much like a C++ file.
> 
>> 3/
>> And, finally, shouldn't .cpp files that contain qdoc comment blocks
>> contain the "right" license for documentation, as it's different from
>> code? Then a rename would also not require a re-licensing.
> 
> Theoretically supercorrect would be a strict separation between "documentation" and "source code". Such an endeavor would be quite costly. I wonder, what problem are we trying to solve here?


What Joerg says.

It might not be perfect in every aspect, but it’s certainly good enough and workable.

Volker