I'm promoting use of the FDL for textbooks and lecture notes at our university. It critically depends on section 3.
The specific question is: *If I publish my own work as a PDF under the FDL in the web, do I need to provide the LaTeX sources?*
Section 3 of the FDL requires publishing a transparent copy if more than 100 opaque copies are published.
Since PDF is opaque, the background questions are
- Does the FDL apply in full to the original author as well? Ie., does the term `copy' in sec 3 denote `piece' (also original) or `REprocution' (which does not cover the original)?
- IF it applies, how many units are published when I put the file online? One? As many as downloads? [1]
- IF more than 100, publishing the pdf without the tex violates the FDL. Yet, since only the copyright holder could pursue the offence, and that's me, for all /practical/ matters I'd be fine? Clearly, I can't recommend to use a license and infringe it: What license would you recommend for those who like the FDL but are unwilling to reveal their LaTeX sources?
Thanks for any insights, hwe
[1] If one publishes as often as there are downloads, the license would encourage taking content offline as soon as 99 are reached -- hardly the intention of a free license. Also, since this timed taking offline is a prohibitive effort, the FDL would discourage putting the document online: It was made for ~20 students, so handing print-outs would be FDL compliant. I don't believe the FDL wants to provide incentives to hide the text offline, either.
On 05/02/2008, hwe hwe@fsfe.org wrote:
I'm promoting use of the FDL for textbooks and lecture notes at our university. It critically depends on section 3. The specific question is: *If I publish my own work as a PDF under the FDL in the web, do I need to provide the LaTeX sources?* Section 3 of the FDL requires publishing a transparent copy if more than 100 opaque copies are published. Since PDF is opaque, the background questions are
- Does the FDL apply in full to the original author as well? Ie.,
does the term `copy' in sec 3 denote `piece' (also original) or `REprocution' (which does not cover the original)?
The license applies to reusers; the author can release what they like under GFDL (or any other license), because they are the copyright owner.
In practical use on Wikimedia sites, it's generally been taken to mean that whatever the reuser receives under GFDL is the transparent copy - e.g., even if the author made a picture in Inkscape, if he releases a rendered PNG under GFDL then that's the thing that's released under GFDL.
I would assume you don't *have* to release the LaTeX sources, any more than you have to release working drafts or research notes. The document you release under GFDL is the document you release under GFDL.
I welcome correction if I'm wrong on this!
- d.
On Tue, 2008-02-05 at 11:17 +0000, David Gerard wrote:
In practical use on Wikimedia sites, it's generally been taken to mean that whatever the reuser receives under GFDL is the transparent copy - e.g., even if the author made a picture in Inkscape, if he releases a rendered PNG under GFDL then that's the thing that's released under GFDL.
The question of transparency is more objective than that; the file format has to be readily amenable to editing. E.g., a PNG with a lot of text is not transparent, even if that's what the author released.
If the original release from the author is not transparent, I think subsequent distributors could fall foul of the opaqueness rules, and would be unable to distribute according to the license. I wouldn't see that as being any different to releasing a binary under the GPL; just because that's what was released doesn't make it "the source" (the requirements in the two cases are very different, though, so maybe not directly comparable).
Cheers,
Alex.
On 05/02/2008, Alex Hudson home@alexhudson.com wrote:
On Tue, 2008-02-05 at 11:17 +0000, David Gerard wrote:
In practical use on Wikimedia sites, it's generally been taken to mean that whatever the reuser receives under GFDL is the transparent copy - e.g., even if the author made a picture in Inkscape, if he releases a rendered PNG under GFDL then that's the thing that's released under GFDL.
The question of transparency is more objective than that; the file format has to be readily amenable to editing. E.g., a PNG with a lot of text is not transparent, even if that's what the author released. If the original release from the author is not transparent, I think subsequent distributors could fall foul of the opaqueness rules, and would be unable to distribute according to the license. I wouldn't see that as being any different to releasing a binary under the GPL; just because that's what was released doesn't make it "the source" (the requirements in the two cases are very different, though, so maybe not directly comparable).
Who could they fall foul of? Remember that the risk model is: "if I use this in a manner outside of 'all rights reserved,' will keeping to the terms of the licence be a sufficiently strong defence?"
If the author released a PNG with text on it and then sues me for making a copy available under GFDL - or even a modified copy under GFDL - I find it hard to imagine a judge doing other than telling the original author not to be silly.
- d.
I'm promoting use of the FDL for textbooks and lecture notes at our university. It critically depends on section 3.
Why are you promoting use of FDL when you don't want your own work to be free? Or, at least for that work you don't want to be free, FDL makes no sense.
The specific question is: *If I publish my own work as a PDF under the FDL in the web, do I need to provide the LaTeX sources?*
As the copyright holder you can do whatever you want, since noone likely has standing to sue you. However, so doing makes the licence worthless, since the rights which it attempts to grant are not effectively granted that way.
- IF it applies, how many units are published when I put the file
online? One? As many as downloads? [1]
My understanding is, as many as downloads.
- IF more than 100, publishing the pdf without the tex violates the
FDL. Yet, since only the copyright holder could pursue the offence, and that's me, for all /practical/ matters I'd be fine?
You'd be fine, but you wouldn't be granting the rights the FDL is designed to grant, so why bother licencing the work under the FDL in the first place? Issuing contradictory licences isn't useful, and it degrades the "brand value" of those licences, if you see what I mean. People who see an FDL-covered document have and should have the expectation that it is effectively free within the meaning of FDL, and applying FDL to documents which are not reduces the confidence people place on such promises.
Clearly, I can't recommend to use a license and infringe it: What license would you recommend for those who like the FDL but are unwilling to reveal their LaTeX sources?
It depends on what is it you like about FDL. Perhaps a Creative Commons attribution, no-derives, would serve your purposes? No-derives since, hiding the source seems to be directed towards that aim, although I'm not sure why you don't want to expose your LaTeX, so can't say.
[1] If one publishes as often as there are downloads, the license would encourage taking content offline as soon as 99 are reached -- hardly the intention of a free license. Also, since this timed taking offline is a prohibitive effort, the FDL would discourage putting the document online: It was made for ~20 students, so handing print-outs would be FDL compliant. I don't believe the FDL wants to provide incentives to hide the text offline, either.
Well, the FDL primarily doesn't want to provide incentives to hide the source.
HTH --David.
First, thanks a lot for the constructive information.
Second, to set the record straight: It's not about texts from me personally, I release the .tex. I just cut "the professor whom I've suggested the FDL for lecture notes" to "I", as it's irrelevant for the answer. (Now `he' and `I' are precise.)
It depends on what is it you like about FDL.
Foremost freedom 0, freedom 1, freedom 2, and freedom 3. Then the copyleft. And the clear delineation of modifications. So I don't think it's fully fair to say he "does not want his work to be free."
Please keep in mind this is about text. Contrary to a program, no information is "hidden" in the source. Anyone who knows LaTeX can easily tell the few \section, \emph or \footnote tags -- the only opaque words.
Even without the mentioned tools to edit pdfs, most of the text can simply be copied and pasted; only with formulae problems are likely. Thus it's not so much a matter of freedom, but of convenience.
No-derives since, hiding the source seems to be directed towards that aim
It's not, also freedom 3 should be included. I fully agree the .tex /should/ be published, to not complicate derivatives. (Frankly, I'm not sure why showing the .tex is an issue. I guess it's just not written as nicely as everything else he shows.)
The problem is that the license choice will likely apply to all works (other than papers to publish), potentially for several professors, and mostly the source /would/ be available. That's why I hesitate to suggest a CC license instead of the FDL. Any recommendations under this circumstances?
Finally, when GNU PDF [1] is done, will pdf qualify as transparent?
Thanks again for plenty of great input, hwe
PS: What if a transparent original never existed? What to suggest to those who still use MS Office? (Other than switching.) (Also, a hand-written manuscript under FDL is not meaningless I guess.)
disappointed by more non-free-software discussion on FSFE's list
@MJ Ray: I've read your opinion on the FDL, and actually in some respects I agree. Yet the FDL is part of the GNU project. If GNU is not to be discussed at FSFE, what is?
On 06-Feb-2008, hwe wrote:
It depends on what is it you like about FDL.
Foremost freedom 0, freedom 1, freedom 2, and freedom 3. Then the copyleft. And the clear delineation of modifications. So I don't think it's fully fair to say he "does not want his work to be free."
If the copyright holder believes freedom 3 to be valuable, this is incompatible with refusing to distribute the source code to recipients. Access to the source is a precondition of freedom 3.
Please keep in mind this is about text. Contrary to a program, no information is "hidden" in the source.
It's just as easy to say that a program, in binary form, has no information "hidden" in the source. A sufficiently determined recipient can always derive an equivalent source that will compile to the same binary form.
This is a bogus argument for programs, and it's a bogus argument for any other software work. Access to the corresponding source code to the work is a precondition to meaningful exercise of freedom 3.
Anyone who knows LaTeX can easily tell the few \section, \emph or \footnote tags -- the only opaque words.
Being able to guess at the source, while not receiving that source, is not satisfying freedom 3. The recipient needs full access to the corresponding source form of the work, not merely the freedom to guess.
Furthermore, if the copyright holder actually believes "no information is hidden in the source", then that leads to the conclusion there is no good reason to hide the source from the recipient.
It's not, also freedom 3 should be included. I fully agree the .tex /should/ be published, to not complicate derivatives. (Frankly, I'm not sure why showing the .tex is an issue. I guess it's just not written as nicely as everything else he shows.)
If you don't understand what's keeping him from satisfying freedom 3, that seems an important avenue of inquiry. You can't recommend a license text to him until you understand what it is he wants.
disappointed by more non-free-software discussion on FSFE's list
@MJ Ray: I've read your opinion on the FDL, and actually in some respects I agree. Yet the FDL is part of the GNU project. If GNU is not to be discussed at FSFE, what is?
I interpreted MJ Ray's comment on "non-free-software discussion" to refer to the fact that you're discussion how to satisfy someone's desire not to reveal the source code to their work (an eminently non-free requirement). I might have interpreted his comment incorrectly.
Foremost freedom 0, freedom 1, freedom 2, and freedom 3. Then the copyleft. And the clear delineation of modifications. So I don't think it's fully fair to say he "does not want his work to be free."
Well, those freedoms require source. Sure, it's possible to edit PDFs, but I think you'd agree that not at the same semantic level at which it is possible to edit .tex source, same thing with source and binaries for software, in principle (and in practice) binary patches are possible.
Please keep in mind this is about text. Contrary to a program, no information is "hidden" in the source. Anyone who knows LaTeX can easily tell the few \section, \emph or \footnote tags -- the only opaque words.
Well, given that LaTeX can do a lot more than sectioning and footnotes, it's not as clear as that, as I see it.
Even without the mentioned tools to edit pdfs, most of the text can simply be copied and pasted; only with formulae problems are likely.
Problems are likely is an understatement here. Unless PDF tools have gotten a lot more adept at capturing semantic mathematical relations than I remember them being.
Thus it's not so much a matter of freedom, but of convenience.
In principle, yes. In practice, freedom is about convenience, too, otherwise no licence would require source-showing, or even go to the extent of forbidding obfuscation (gpl3).
It's not, also freedom 3 should be included. I fully agree the .tex /should/ be published, to not complicate derivatives. (Frankly, I'm not sure why showing the .tex is an issue. I guess it's just not written as nicely as everything else he shows.)
Right. So we mostly agree. Personally I'd say that, unless this is very high profile, it's not that likely that people will be fiddling around with the sources, and if they do, they'll probably know that sources aren't necessarily things of perfection ;-)
The problem is that the license choice will likely apply to all works (other than papers to publish), potentially for several professors, and mostly the source /would/ be available. That's why I hesitate to suggest a CC license instead of the FDL. Any recommendations under this circumstances?
Creative Commons, Attribution, Share-alike?
Finally, when GNU PDF [1] is done, will pdf qualify as transparent?
I'm not convinced of that. LaTeX compilation loses a very significant amount of semantic information. Even if PDF can be parsed and well understood in its own terms, editting characters on a canvas is very different from, say, changing \parskip
HTH, --David.
On 07/02/2008, David Picón Álvarez eleuteri@myrealbox.com wrote:
I'm not convinced of that. LaTeX compilation loses a very significant amount of semantic information. Even if PDF can be parsed and well understood in its own terms, editting characters on a canvas is very different from, say, changing \parskip
Is it nevertheless the case, though, that what is released under GFDL is what is released under GFDL, and if you get a PDF or a bitmap or whatever then, as a reuser, you're not somehow obligated to release a more transparent version than you got from the copyright owner? (Given the author can't reasonably sue you for copyright violation for reusing precisely what they gave you rather than what they didn't.) This is quite an important question for Wikimedia content!
- d.
Is it nevertheless the case, though, that what is released under GFDL is what is released under GFDL, and if you get a PDF or a bitmap or whatever then, as a reuser, you're not somehow obligated to release a more transparent version than you got from the copyright owner? (Given the author can't reasonably sue you for copyright violation for reusing precisely what they gave you rather than what they didn't.)
I'm not going to stick my neck on this one, I have no idea. FWIW maybe this is one of those cases in which, if you can't reuse in accordance with the licence, you cannot distribute at all (liberty or death clause).
--David.
* David Picón Álvarez eleuteri@myrealbox.com [2008-02-07 12:36:05 +0100]:
Foremost freedom 0, freedom 1, freedom 2, and freedom 3. Then the copyleft. And the clear delineation of modifications. So I don't think it's fully fair to say he "does not want his work to be free."
Well, those freedoms require source. Sure, it's possible to edit PDFs, but I think you'd agree that not at the same semantic level at which it is possible to edit .tex source, same thing with source and binaries for software, in principle (and in practice) binary patches are possible.
I think it would be better if the students get the LaTeX code, than they can also learn how to make those nice documents. Or they can be told, that the paper was written under time pressure and therefor not very nice written. Patches to the tex file could get a bonus :) (e.g. put the the stuff under version control.)
Beside that PDF is listed as example for a transparent copy:
Examples of suitable formats for Transparent copies include plain ASCII without markup, Texinfo input format, LaTeX input format, SGML or XML using a publicly available DTD, and standard-conforming simple HTML, PostScript or PDF designed for human modification. (http://www.gnu.org/licenses/fdl.html)
Best wishes, Matthias
I think it would be better if the students get the LaTeX code
Of course it would! The question is *not* "should the tex come with an FDL-pdf?" We all agree on that.
The question is: Given a dozen of simply formatted text documents, given they'll all carry the same license, given most of them will but a few will not come with the .tex, given the author wants anyone to have the 4 freedoms (as conveniently as possible under this restrictions), is it better to put "All rights reserved" (status quo), a CC license or the FDL?
I think we've established legally FDL is possible. The remaining controversy is about what is better morally.
One position is "if it's not ultimately free, dismiss it." (And some count FDL as *non*free.) I rather support "make it as free as possible. Then keep improving."
Besides, let's not forget that handing out printouts is perfectly fine under FDL, since they're surely fewer than 100. (Anyone who thinks this is wrong should comment the drafts for FDL2)
Beside that PDF is listed as example for a transparent copy:
This would settle the issue! Thanks for pointing it out, I looked the FDL up in RMS' book and there's only ver 1.1.
However, will pdflatex resp. dvi>ps>pdf give "PDF designed for human modification" or "PDF produced by some word processors for output purposes only"?
Thanks again, hwe
PS: What about the no-transparent-original question -- if its pdf counts as opaque, eg. PowerPoint presentations can't be FDL'ed (unless there're less than 100 copies).
On 08/02/2008, hwe hwe@fsfe.org wrote:
One position is "if it's not ultimately free, dismiss it." (And some count FDL as *non*free.) I rather support "make it as free as possible. Then keep improving."
GFDL is free by all measures if you don't use invariant sections and so forth, isn't it? Does anyone dispute that? (I know Debian wouldn't accept it with invariant sections.)
- d.
One position is "if it's not ultimately free, dismiss it." (And some count FDL as *non*free.) I rather support "make it as free as possible. Then keep improving."
GFDL is free by all measures if you don't use invariant sections and so forth, isn't it? Does anyone dispute that? (I know Debian wouldn't accept it with invariant sections.)
A document licensed under the GFDL with invariant section is still free, what it isn't though is free software; but it is free documentation.
On 08-Feb-2008, David Gerard wrote:
GFDL is free by all measures if you don't use invariant sections and so forth, isn't it? Does anyone dispute that?
Unmodifiable sections (not merely those the license calls "Invariant Sections") are only the most prominent of the problematic issues of the FDL. Even if a specific work does not exercise those parts of the license, the other problems remain.
"Draft Debian Position Statement about the GNU Free Documentation License" Not ratified, but does cover many of the problems with the freedom of works under the FDL. URL:http://people.debian.org/~srivasta/Position_Statement.xhtml
"A Simple Guide to the Problems of the GNU FDL" URL:http://mjr.towers.org.uk/blog/2006/fdl
Many of the remaining problems have to do with the license attempting to have "documentation" distinct from "program", even though there are many works that are clearly both (e.g. Postscript documents).
It's unfortunate that some within the FSF choose to interpret "software" as equivalent to "program", instead of the more tenable position that "software" is a term as opposed to "hardware". This leads to the even worse position that documentation recipients deserve freedoms different from the freedoms deserved by recipients of programs.
The question is: Given a dozen of simply formatted text documents, given they'll all carry the same license, given most of them will but a few will not come with the .tex, given the author wants anyone to have the 4 freedoms (as conveniently as possible under this restrictions), is it better to put "All rights reserved" (status quo), a CC license or the FDL?
`a CC license' includes a class of non-free documentation licenses that for example prohibit you from using them commercially, and other nasty things. `All rights reserved' is also a non-free documentation license as well, so the only choice is really the GNU Free Documentation License.
One position is "if it's not ultimately free, dismiss it." (And some count FDL as *non*free.) I rather support "make it as free as possible. Then keep improving."
They count it as a non-free _software_ license, which it is true, since it isn't a software license to begin with; it is a documentation license, and a free one at that.
Cheers!
On 05-Feb-2008, hwe wrote:
What license would you recommend for those who like the FDL but are unwilling to reveal their LaTeX sources?
I would recommend a psychotherapy session first. How can one "like the FDL" but simultaneously not want to give recipients the source to the software? That's fundamental to the goals of the license.
Clearly, the goals of these people are not compatible with freedom of the work.
[1] If one publishes as often as there are downloads, the license would encourage taking content offline as soon as 99 are reached
Rather, the license encourages publishing source in all cases, to avoid exactly the issue you're describing.
hardly the intention of a free license.
It's also hardly the intention of a free license to permit restricting the freedom of recipients of the work. That seems to be at odds with the goals of the people you represent.
Ben Finney ben@benfinney.id.au wrote:
On 05-Feb-2008, hwe wrote:
What license would you recommend for those who like the FDL but are unwilling to reveal their LaTeX sources?
I would recommend a psychotherapy session first. [...]
I'm not going to repeat the problems with FDL in full http://mjr.towers.org.uk/blog/2006/fdl#general and I'm disappointed by more non-free-software discussion on FSFE's list, but I would point out that usually you cannot convince an insane person to undergo psychotherapy by telling them they need one.
Show the absurdities of not revealing the source. Then encourage them to reveal their LaTeX sources, but shackle it to unremovable adverts and poison pills, in line with FDL's goals. I'd expect over 90% of academic publications would be happy as adware.
Regards,