Re: Font license recommendation
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:
>> At 00.53 +0200 2002-08-03, Thomas Bushnell, BSG wrote:
>> >Since things like intention matter--and not just technical
>> >mechanism--this is just FUD.
>>
>> "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.
>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.
>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. _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.
>> 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.
>> 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.
>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.
>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. 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.
>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. The preferred format
for modifying the Computer Modern fonts is clearly the MF sources, even
though dvips by default embeds them as bitmaps in PS output and most DVI
viewers also rely on bitmaps for displaying documents on screen. Similarly,
a font created as a Type1 or TrueType has that as preferred format, even if
it (for some bizarre reason) was to be embedded into a PS file as bitmaps.
>at that point, what would it matter if the fonts were embedded in another
>file?
>
>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?
Lars Hellström
Reply to: