Re: Font license recommendation
On Thu, 2002-08-08 at 14:30, Lars Hellström wrote:
> On 04 Aug 2002 20:22:11 -0500, Jeff Licquia <licquia@debian.org> wrote:
> >On Sun, 2002-08-04 at 17:53, Lars Hellström wrote:
> >> "FUD" ?
> >>
> >> On what do you base your opinion that intent has any significance for
> >> whether the GPL allows an action?
> >
> >Well, there's this:
> >
> >"The source code for a work means the preferred form of the work for
> >making modifications to it." (section 3)
> >
> >"Preferred" is an intent, last time I checked.
>
> Intent. A purpose or aim; meaning; import. Syn.: Design, purpose,
> intention, drift, significance, view, aim.
>
> Prefer. To regard or esteem more than something else; to place in a
> higher rank or position; to give priority to; to seek recompense.
>
> In a broad sense of intent perhaps, but I got clear impression Thomas had
> something much more concrete and close in mind.
Maybe, maybe not.
> >It's used against people
> >who use obfuscators on their source before distribution, for example,
> >because no one in their right mind would intend to edit the obfuscated
> >source to add features or fix bugs.
>
> True, and as I mentioned (see below) there is a lot in the license about
> the intent of the license, but the question was rather about whether the
> intent of the person who performs an action has any significance for the
> GPL allowing that. If you insist on that intent matters, then you have to
> explain why it could be OK to link GPL-compatible and GPL-incompatible code
> into one program when
> (i) the person doing it wants X, but not when he wants Y; or
> (ii) the program does X, but not when it does Y.
Last time I checked, licenses don't have the power to "prefer" anything,
to modify code, or to run obfuscators. "Intent" is a trait held by
people; the fact that we use handy linguistic shortcuts does not any
more give licenses an "intent" than it gives seafaring vessels sexual
characteristics by being called "she". Licenses are statements of
intent by the licensor; they have no intent of their own.
It's always OK to link GPL-compatible and GPL-incompatible code, no
matter what the intent of either the licensor or the licensee. But if
the person doing the linking distributes the result, that's not OK.
Again, the intent of either party is not at issue. The question is
whether the license on some bit of code is incompatible with the GPL,
and to answer that question, intent has to be taken into account.
> >There's also this:
> >
> >"The act of running the Program is not restricted, and the output from
> >the Program is covered only if its contents constitute a work based on
> >the Program (independent of having been made by running the Program).
> >Whether that is true depends on what the Program does."
> >
> >"What the Program does" is also intent.
> >
> >Incidentally, the latter paragraph also provides reason to believe that
> >a PostScript file generated using a GPL font does not become "tainted"
> >with the source requirements of the GPL. (No, it's not clear-cut; I'm
> >not necessarily arguing that it's the case, just that it's possible.)
>
> But the PS file is no more output from the font than a binary executable is
> output from an object code file which was linked into it. The font is not
> "run" until the document is rendered.
I'm assuming that the font is rendered into bitmaps. This wasn't clear,
I'm afraid.
> _Maybe_ you could make an argument
> that distilling a PS file is running it and that fonts in the PDF are
> output of the fonts in the PS, but that is shaky too. In particular, the
> procedures for actually drawing the glyphs (which are what commercial
> copyright holders are most keen on protecting) are generally left unchanged.
Compilers generally don't change during the compile process, either.
Does that change the fact that the object files they product are "output
of the Program" in the GPL sense?
> >> It clearly says
> >>
> >> You may not copy, modify, sublicense, or distribute the Program
> >> except as expressly provided under this License. Any attempt
> >> otherwise to copy, modify, sublicense or distribute the Program is
> >> void, and will automatically terminate your rights under this License.
> >>
> >> and I don't see any reference in it to the intent of the licensee (only to
> >> the intent of the license, but that is something quite different).
> >
> >That's because intent is a subject of other parts of the license.
>
> Both "it"s refer to the license as a whole.
...which expressly refer to the intent of various parties.
> >> Furthermore I don't see that there would necessarily be any difference in
> >> intent. Certainly if one writes a program whose only purpose is to
> >> demonstrate a legal loophole there would be a difference in intent, but
> >> that isn't the interesting case. In the interesting case the intent is to
> >> "make a single file program, incorporating various pieces of free software
> >> (some of which are GPL and some of which are GPL-incompatible), that does
> >> X". The "X" could be to display a certain picture; PS files can have this
> >> intent, but there are also C programs with the same intent.
> >
> >Well, let's take the Gimp. It processes one image file into another,
> >and is GPLed. Does that have any implication on the legal status of
> >either image? What if the original is GPLed? Especially given that
> >most image formats are their own preferred format for modification,
> >things are not so clear as they might seem.
>
> Gimp requires very precise input to produces the picture, which is why the
> GPL paragraph you quoted above is applicable; the output is not a work
> based on Gimp. Likewise, when I use a postscript printer driver, I don't
> claim that the PS file it produces is a work based on the driver.
I don't recall that "precision" is a condition recognized by the GPL.
Perhaps you could quote me the relevant paragraph?
> >One could make the case that a font is an interpreted language, with the
> >font renderer as the interpreter.
>
> _Postscript_ is very much an interpreted language, PS fonts are code
> libraries (with exception for pure bitmap fonts, which are rather data, but
> those are rare these days), and GhostScript is an example of an interpreter
> for that language.
>
> >In that case, a rendered glyph could
> >be considered "output from the Program".
>
> Agreed.
>
> >OTOH, the font could be
> >considered data that is input by a font renderer; in that case, the
> >rendered glyphs could be considered "derived works".
>
> Derived works of what? The renderer? That would be like saying GhostScript
> is derivative work of gcc, and then we're very much back in the problem
> with tainted software.
No, of the fonts themselves. If the font is data that is processed by a
font renderer to produce glyphs, then the glyphs are derived works of
the fonts. Moreover, by looking at the question in this way, we are
denying that the glyphs are "the output of the Program" (the Program
being the fonts), which means that this exception in the GPL would not
apply.
> >Then again, if you're using the Gimp to create a PNG with text rendered
> >in a GPLed font, it wouldn't matter; the PNG file is its own "preferred
> >format".
>
> Not necessarily. The PNG could be a blueprint for some piece of machinery,
> originally drawn in a CAD program, and then the file format of that program
> would certainly be the preferred format for making changes.
Right. I originally said that I was "using the Gimp". CAD drawings are
equivalent to EPS/SVG files for the purposes of this discussion.
> I maintain
> however that this PNG (a bitmap, right?) is "output from the program" (both
> regarding Gimp and regarding the font), and that it therefore is the input
> controlling Gimp that determines the status of the PNG.
For the Gimp, yes. But the font? Remember, it's not totally clear that
fonts are programs or data. If they aren't programs, then the PNG can't
be their output, which makes them derived works under the GPL's
conditions.
> >But an EPS or SVG might, especially if the font is required to
> >process the file properly. Indeed, a vector image file format could
> >itself be described as a program; PostScript is even considered Turing
> >complete, if my memory serves me.
>
> Not too familiar with the technical details, are you? Postscript is (modulo
> implementation limits) as powerful as Lisp. :-)
>
> >But what if you convert a PNG to an
> >EPS, and add some text with a GPLed font? What constitutes "the
> >source"? If the PNG is proprietary, have you just violated the GPL?
> >
> >The details of embedding could even make a difference.
>
> Yes. However if the details of embedding could make a difference for a font
> in a PS file, then they should also make a difference for a library in an
> executable.
>
> >One could embed
> >bitmaps generated by the font instead of the font itself, which would
> >clearly be "output from the Program".
>
> Yes, but that creates huge problems with output quality. People wants to
> avoid it whenever possible.
>
> >Bitmap fonts would themselves be
> >in the "preferred form", so their licensing status might not matter;
>
> Only fonts that were originally created as bitmaps.
That's what I meant.
> >In short, I don't think the question is as simple as you make it.
>
> That's funny. I objected to Thomas's remark mainly because IMHO the matter
> isn't as simple as he made it. Now what was the question that you don't
> think is simple?
This point:
"And just to make sure we're clear on what my point is: Incorporating a
GPLed font in a PS document does, in contrast to what you claimed, have (in
many cases) unwanted legal implications; if it didn't then there would be a
simple workaround for GPL-incompatibility."
You claim it has unwanted legal implications; I claim that the question
isn't settled by a long shot.
Reply to: