Re: Bug#383481: Must source code be easy to understand to fall under DFSG?
Francesco Poli wrote:
> On Thu, 02 Nov 2006 18:12:05 -0600 Terry Hancock wrote:
>> Yes, it seems the question here is *whose* "preferred form for
>> The GPL appears to assume there is a general consensus on this, and
>> there may not be.
> IMO, whenever there's any doubt about which is the preferred form
> ("preferred by whom?"), we should follow the preferences of the last
> modifier: if you actually modify a work, you've shown in practice
> what is your preferred form for modifications (rather than simply
> claiming in a vacuum what *would* be your preferred form, should you
> make modifications that you don't actually make!).
While I agree with that conceptually, it may impossible to know. The
person making the call is going to be a package maintainer (subject to
review by debian-legal and ultimately the ftp masters), but it may not
be easy to tell whether the upstream source has included the source as
they modified it, or merely source that could be modified.
This is likely to bring up things like CASE tools, project files, and
change control systems. How much of these things are merely "tools for
manipulating the source" and how much are "part of the source"?
I don't think there's a hard line that can be followed in all cases, so
a judgement has to be made by somebody in deciding what to include and
what not to include.
> That is the interpretation that allows source to change form,
> whenever someone needs so. Hence, if Fred Fortraner writes a program
> in Fortran77 and releases it under the GNU GPL v2, David Distributer
> must make the (unobfuscated) Fortran77 code available as source. But
> then, Pete Pythonhacker can take the program, translate it into
> Python (because he cannot stand Fortran) and go on modifying the
> Python derivative: at that point, the source for the derivative is
> Python code... And so forth.
Yes, that makes sense. But in most cases we can't really tell whether a
conversion was made by an automatic process or by the author.
>> It's clearly going to be a judgement call, but I'd guess that "in a
>> standard format" which will be understood by potential modifiers
>> is more important than "in the original form" which may not be.
> I disagree: there's no "standard format" (whatever that may mean)
> requirement in the GPL definition of source code.
This is not so much of an issue for program source code as it is for
other types of work, but what I mean is that an SVG is probably better
than an XCF (more universal editor support). Even though Larry Ewing
originally drew his Tux in Corel Draw (IIRC), it's the SVG or Skencil
formatted versions that are better to include in Debian (because we have
free software tools to edit these formats).
I guess what I'm saying is that in these cases it's more obvious which
form will be of greater use to would-be modifiers who use the
distribution, whereas the upstream source may be happier with Corel Draw.
It's more important that the end user is empowered to exercise his
"freedom to modify" than that the historical process of creation has
been preserved (one could pick endlessly over the details of the latter,
but the former is what actually gives meaning to free licensing of a
work). So if there's a discrepancy between what is going to be easy for
the Debian user and what happened to be easy for the upstream author,
then the user's PoV ought to be the one that really matters.
>> Of course in your specific case, I think it'd be better to have
>> your original XML *and* the XSLT translator so that the HTML can be
>> generated from it. The question, I would suppose though is not one
>> of "best practice" but of "permissable practice", and I think the
>> HTML is good enough.
> I don't think so. Machine-generated HTML is not source, at least not
> until it becomes the form that is actually preferred for making
> modifications to the work...
I think you have to recognize some kind of pragmatic limit to these
kinds of restrictions. The HTML certainly can be modified, and a lot
more people know HTML than some ad hoc XML format. It may well be the
case that the HTML would be the preferred form for modification for
anyone but the original author.
Terry Hancock (hancock@AnansiSpaceworks.com)
Anansi Spaceworks http://www.AnansiSpaceworks.com