|
|
Subscribe / Log in / New account

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Bruce Perens wonders how many open source licences we really need. "The Open Source initiative has, to date, approved 73 licenses. How many do you really need? If you're a company or individual producing Open Source software, no more than 4. And you can get along with just 2 of them."

(Log in to post comments)

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 16, 2009 22:00 UTC (Mon) by mheily (subscriber, #27123) [Link]

This article has a nice summary of the issues around choosing a free software license, but I find the scare-mongering around the issue of license proliferation a bit too much to take. For example, it states:

> [T]he fact that there are 73 licenses is a problem. Many of those licenses are incompatible with each other. To understand the legal implications of mixing software under two of those licenses together in the same program, you'd have to learn 5256 different combinations!

While Bruce is technically correct about the number of licenses, according to the OSI, only nine of the 73 licenses are "popular and actively used" [1]. The rest are special purpose, non-reusable, obsolete, deprecated, etc., that would not be considered for any new software being developed.

> So, [use either Apache 2.0 or one of the GPLv3 licenses]. Suddenly, Open Source isn't as complicated as those 5,256 combinations of two of the approved 73 licenses!

So, basically the entire article is an attempt to scare people into using the GPLv3 (or the Apache 2.0 license, if you are forced to by an employer). If you don't, get ready to spend years studying thousands of license combinations :)

[1] http://www.opensource.org/minutes20060712

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 16, 2009 23:49 UTC (Mon) by BrucePerens (guest, #2510) [Link]

While you might not meet all of those licenses, I assure you that those of us who have worked on creating Linux distributions that can be legally distributed meet every one. And more. This is why it was essential for Debian to have the DFSG. I couldn't tell people what was OK to include and what was not otherwise.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 17, 2009 19:08 UTC (Tue) by nevyn (subscriber, #33129) [Link]

While Bruce is technically correct about the number of licenses, according to the OSI, only nine of the 73 licenses are "popular and actively used" [1]. The rest are special purpose, non-reusable, obsolete, deprecated, etc., that would not be considered for any new software being developed.

While the article by Bruce has some flaws, I spoke to the Fedora license person very recently and he said that the number of "Open Source" licenses is increasing. Apparently a significant portion of these are "I'm not a lawyer and re-wrote the MIT license, using my own words" but still, it's worry and not a good trend.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 17, 2009 5:44 UTC (Tue) by dhess (guest, #7827) [Link]

Hi Bruce, thanks for another informative essay.

For my own work, I used to choose GPLv2 by default, but lately I've switched to MIT, because in most cases what I'm releasing has marginal utility, "works for me," etc. I assume there's little chance that a third party will use my code, let alone redistribute a proprietary fork, and even if someone did, I wouldn't particularly care. It is, as you put it so well, a "gift" (maybe not a very special or interesting gift, but that's beside the point ;). I get more value from knowing somebody else might use it than I would get in the unlikely event of some third-party improvements to the code.

When I participated in the release of the OpenEXR library, one of the goals of the release was to encourage the use of the code in proprietary software such as Renderman or Photoshop. We figured that the risk of a proprietary fork was mitigated by the network effects of the widespread use of our version of the library, so it was in our best interests to eliminate any controversy over the license. There are multiple precedents for MIT/BSD licensed-software in proprietary systems (libjpeg and libpng, to name a couple); we chose the modified BSD license for its advertising clause. I think that turned out to be a good choice, judging by the library's uptake.

In either circumstance -- "care about proprietary forks," or not -- what they have in common is that I don't want the license to discourage use. It should be a no-brainer. Considered in that context, I perceive both the familiarity and conciseness of the BSD and MIT licenses to be advantages over the Apache 2.0 license. That license is arguably less familiar to most and obviously more verbose. Speaking for myself, when I see a free software package that uses MIT or BSD, I instantly have a sense of its licensing implications for my program. I can't say the same for the Apache 2.0 license, due to its more complicated legalese and less common usage [1]. And although I haven't looked carefully, I'm not aware of any proprietary software vendors using a third party's Apache 2.0-licensed code in their products. (Maybe Google? But then, they're not a typical vendor.)

Of course, I also don't want to be sued, and you allude to greater legal protection in your in your recommendation of Apache 2.0 over other "gift licenses." However, I'm not aware of it ever being tested in court, so I'm uncertain of the worth the extra verbiage.

So when I compare Apache 2.0 to BSD or MIT for either of the purposes I described above, I see a couple of drawbacks in choosing Apache 2.0, and one theoretical advantage. I'd like to participate in license non-proliferation, but I'm not yet comfortable with the license you recommend. What are your thoughts on these tradeoffs? Do you know of any cases where the Apache 2.0 license has protected a free software developer from patent claims? Have you talked to any large proprietary software companies about their comfort level with third-party Apache 2.0-licensed code?

---

[1] Hardly a scientific study, but my quick check of licenses used by Sourceforge projects revealed that BSD and MIT licensed-projects outnumber Apache 2.0 by a ratio of over three to one. (Sourceforge's interface doesn't allow me to distinguish BSD from modified BSD.) I couldn't figure out any quick way to break down licenses for projects hosted at Savannah, Gitorious or GitHub, sadly.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 17, 2009 5:51 UTC (Tue) by dhess (guest, #7827) [Link]

Ugh, I really botched the 2nd-to-last paragraph. It should read:
Of course, I also don't want to be sued, and you allude to greater legal protection in your recommendation of Apache 2.0 over other "gift licenses." However, I'm not aware of it ever being tested in court, so I'm uncertain of the worth of the extra verbiage.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 17, 2009 12:45 UTC (Tue) by nim-nim (subscriber, #34454) [Link]

While pure software licensing is well understood and people complain of too many licenses many other digital works do not have really satisfying authoritative licenses (for example, fonts, every single recent high-profile font release such as Vera, Liberation, Droid, Stix used ad-hoc licensing).

Since those works are also necessary to get a satisfying FLOSS system (it's not *all* software), are just as difficult to re-license later, and have long lifetimes (decades is not uncommon, see TEX), we'll pay for the software-only focus of FLOSS lawyers later.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 17, 2009 16:13 UTC (Tue) by BrucePerens (guest, #2510) [Link]

Unfortunately, the last two attempts at producing font licensing submitted to OSI for approval have been giant steps backward compared to the software licenses. One had a loophole big enough to drive a truck through that would allow anyone to convert the font to public domain. One doesn't allow format conversion or the distribution of modified versions as font files.

The first was written by software developers without the assistance of an attorney. They should know better. The second comes from a government project that's still coping with fear of modification.

Bruce

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 17, 2009 16:30 UTC (Tue) by nim-nim (subscriber, #34454) [Link]

That just shows there is a need to look seriously as non-software licenses, because they're not here yet.

Till it's done bad license proliferation will continue and good luck trying to convince Ascender or Bitstream to re-license fonts they've released under one of those later, once their contracts have expired and there is no money in it for them anymore.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 6:33 UTC (Wed) by njs (subscriber, #40338) [Link]

I had the vague impression that the SIL OFL was the current best font license, and they claim prominently to be OSI-compatible (though not, perhaps, OSI-approved):
http://scripts.sil.org/cms/scripts/page.php?site_id=nrsi&...

So I'm curious, is it one of the license you mention, or has it just not been submitted for OSI approval, or...?

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 7:05 UTC (Wed) by BrucePerens (guest, #2510) [Link]

Yes. It was written without the assistance of an attorney, and that really shows, and it has little chance of actually working in a court the way they expect it to work. Their response to criticism of the poor construction is "the developers wanted it that way". There is a loophole in the license that allows you to convert any font under it to public domain. No kidding. There is additional language of very questionable meaning.

Don't use it. Don't let your friends use it.

Bruce

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 7:36 UTC (Wed) by nim-nim (subscriber, #34454) [Link]

I wouldn't go that far because the *other* licenses are usually worse for fonts.

Which brings me back to my first point: some domains do not have too many good licenses to choose from, they don't have one clearly good license.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 8:13 UTC (Wed) by BrucePerens (guest, #2510) [Link]

There is no license I know of worse than the SIL license. If you really want to use it, please just put your work in the public domain. Because that way you won't be fooling yourself about what rights the license gives you.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 9:14 UTC (Wed) by nim-nim (subscriber, #34454) [Link]

Again, this is an overstatement.

It's trivial to find worse licenses just look at any font in your distribution and check its license text.

Most of the GPL fonts can not be safely used with PDFs or any document format that embeds fonts. But this is what a lot of people use fonts for! This is a worse problem that PD conversion (let's be honest our fonts are not at the state people line up to misappropriate them, and those who do usually ignore any licensing; also PD makes it possible to actually relicence fonts later on)

The non GPL-fonts either use the OFL or homegrown licenses which have even more artisanal wordings.

It would be great if the OSI or the FSF which obviously have access to competent attorneys proposed one solid FLOSS font license. But till they do, well for all the OFL faults I find it hard to point the finger at OFL proponents, since there are not real alternatives.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 9:30 UTC (Wed) by BrucePerens (guest, #2510) [Link]

It's trivial to find worse licenses just look at any font in your distribution and check its license text.
I am pretty familiar with the licenses in my distribution, having written the rules my distribution uses to accept them. No, no license text quite that bad elsewhere in the distribution.

The question of GPL fonts is best addressed in David Turner's discussion of the Font Exception. He addressed, in 2005, the very same point that the SIL guys got wrong later on. He had about the best legal counsel around in Eben Moglen.

Bruce

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 9:42 UTC (Wed) by nim-nim (subscriber, #34454) [Link]

Either you don't ship Texlive or you haven't looked at it closely then.

The GPL font exception is nice, but any reality check will show the majority of GPL fonts does not use it.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 16:37 UTC (Wed) by BrucePerens (guest, #2510) [Link]

Debian does not include some parts of TexLive that have licensing issues. Much of TexLive duplicates other packages in the Debian system, and where that is true they've generally used the other sources rather than TexLive.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 17:04 UTC (Wed) by nim-nim (subscriber, #34454) [Link]

Debian ships GUST tetex fonts under the "GUST-FONT-NOSOURCE-LICENSE.txt". It ships the Magenta fonts under the MGOpen license ("or if the modifications are accepted for inclusion in the Font Software itself by the each appointed Administrator"). It ships OCR-A (derived from CTAN's "Free for use but distribution for profit only by arrangement"). It ships the dustin fonts under the GPL (no font exception).

Took me 5 min to collect this, it's not an exhaustive list.

While none of those are exactly non-free (except perhaps OCR-A) claiming they are "better" licensed than if they were OFL-ed is a big stretch.

(and this is not to say bad things about Debian, many distributions are much worse, this is just to show there are many worse things than OFLed fonts out there)

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 17:51 UTC (Wed) by BrucePerens (guest, #2510) [Link]

The version of OCR-A in Debian is in the public domain.

GUST deprecated the no-source license in 2006. But I don't see huge loopholes in it or anything to make me think it's worse.

The MGOpen license states how the license terminates incorrectly:

This License becomes null and void to the extent applicable to Fonts or Font Software that has been modified and is distributed under the "MgOpen" name.
This should be "terminates", not "becomes null and void". But it's a minor problem.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 17:54 UTC (Wed) by BrucePerens (guest, #2510) [Link]

This leaves us with the GPL without font exception. This is arguably problematic for embedded fonts, but only if you can get a judge to agree that inclusion of a font makes the document a derived work of the font, which is a stretch. However, I would suggest that later versions get the exception, for legal clarity.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 18:26 UTC (Wed) by nim-nim (subscriber, #34454) [Link]

With the no-font exception GPL your users are at risk doing just normal tasks like producing a PDF. With the OFL there is a potential risk someone will misappropriate the font (which is at worst what BSD licenses allow, and no foundry has expressed any interest in doing so so far).

IMHO it's clear the first one is the worst license.

Also I don't think anyone ever checked how the FSF font exception works with @font, or with GPLv3.

So, while I'm not a huge fan of the OFL, I don't think "choose any license but the OFL" is a good idea either. And the FSF admits itself its font exception needs more research (not to mention very few projects use it at all because it's not in the default licensing text).

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 18:16 UTC (Wed) by nim-nim (subscriber, #34454) [Link]

> The version of OCR-A in Debian is in the public domain.

It is derived from the CTAN files as documented in wikidepia and the readme of the upstream project. I quoted CTAN's licensing information fully.

> GUST deprecated the no-source license in 2006

But to the best of my knowledge some fonts in Debian have never been re-licensed under something else.

> But I don't see huge loopholes in it or anything to make me think it's worse.

I don't see anything to make me think it's better. It's a russian doll license which is a terrible choice. It has the GPL 2 "or later" problem explicitelly in the licence text

The Mgopen license is fully asymetric and unfriendly to derivatives.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 18:40 UTC (Wed) by BrucePerens (guest, #2510) [Link]

There are two versions of OCR-A in Debian. ttf-ocr-a is in the public domain.
The OCR-A font was developed by the American National Standards Institute (ANSI)
to be readable by the computers of the 1960s.  The OCR-A font is still used
commercially in payment advice forms so that a lockbox company can determine 
the account number and amount owed on a bill when processing a payment.  
A site license for the OCR-A font is very expensive, so I undertook to create 
a free font.  I started with the MetaFont definitions, used FontForge and potrace 
to construct a TrueType font, then assigned each glyph a Unicode code point.  
I visually compared the resulting character shapes with ANSI X3.17-1977.

To use the OCRA font with Microsoft Windows, download file OCRA.ttf and drop 
it in your fonts folder.  To print with it, choose font OCRA and size 10 point.

The other files in this project will be of interest to those who
wish to modify the font.  The shape of each glyph was defined by ANSI as
described in their document ANSI X3.17-1977.  Those shapes were coded
in the MetaFont language as strokes by Tor Lillqvist and Richard B. Wales.
Their work is in the .MF files under MetaFont Sources.  The MetaFont program
and the OCR font definitions are available as part of the TeX package from 
the CTAN archive.

I compiled the OCR-A font into a large bitmap to make edge finding easier; 
that file is ocr10.pk.  I then loaded ocr10.pk into FontForge, which is
available on SourceForge, as background images.  From FontForge I used
potrace, also available on SourceForge, to trace outlines around the bitmaps.
I then assigned a Unicode code point to each glyph.  Mostly they were
obvious, but Unicode does not make provision for the "alternate glyphs"
of ANSI X3.17-1977, so I found some reasonable places to put them.
If Unicode ever adds the "alternate glyphs" of ANSI X3.17-1977 to
their OCR page, the assignments should be updated.  The resulting file
is OCRA.sdf, which can be read by FontForge if you want to modify the
character shapes or change the Unicode code points.

ANSI specifies a character spacing of between 0.09 and 0.18 inch.
I chose 0.1 inch, since that seems to be a common spacing.

FontForge is able to write both PostScript and TrueType font files, 
so I wrote both.  OCRA.afm, OCRA.pfa and OCRA.pfb are the PostScript
font files, and OCRA.ttf is the TrueType font file.  Windows users need
only be concerned with the OCRA.ttf file.  Simply drop it into the fonts
directory and they can print using the OCR-A font by selecting 10-point
size.

The American National Standards Institute makes the X3.17-1977 document
available on their web site for a modest fee.  I purchased that document
so I could compare the font I am publishing with the official definition.
I found an exact match, which speaks well for the efforts of Tor Lillqvist
and Richard B. Wales, since I made no manual changes in the bitmaps or
the character shapes constructed by potrace.  I would like to have included
the X3.17-1977 document in this project, for completeness, but ANSI has
copyrighted it so I cannot.

If you have any difficulty using this font please contact me, I will be
happy to assist you.
    John Sauter (J_Sauter@Empire.Net)
    October 12, 2004

This leaves us with the texlive version. which does seem to be of unsure provenance. If it's not derived from Sauter's version, they should replace it with one that is.

Even if the more recent Debian packages are not consistent with the later GUST license (and I've not checked the Valentine's Day release), the GUST license isn't harmful.

A "Russian Doll" license? "GPL2 or later" is not a problem because you can always use GPL2 if you don't like a later version. Or you can use any one of many later versions. It is important to have the possibility to relicense the work, because some legal problem in an earlier license could make a later one necessary.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 19:19 UTC (Wed) by BrucePerens (guest, #2510) [Link]

Um, let me amend that. Sauter has created his own font program, and the bitmap he used as an intermediary isn't copyrightable, so his version is in the public domain. He seems to have found who made the metafont program, which texlive listed as "author unknown". So, I'm not sure there is a problem with the metafont program. But if there is, a new one can be produced from Sauer's program. This would be circular, but due to oddness in how fonts can be copyrighted, it would work.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 19:38 UTC (Wed) by nim-nim (subscriber, #34454) [Link]

It would work *in the USA*.
That does not help users in other countries.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 19:43 UTC (Wed) by BrucePerens (guest, #2510) [Link]

OK, that's a good point. Let's go back to what's wrong with the CTAN version. Is it more of a problem than just that the archive lost the attribution?

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 20:02 UTC (Wed) by nim-nim (subscriber, #34454) [Link]

CTAN says "Free for use but distribution for profit only by arrangement"

Looks non-commercial to me

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 21:06 UTC (Wed) by nim-nim (subscriber, #34454) [Link]

And at best freeware, not libreware, so…

But anyway my aim was not to dissert on how bad the existing licenses are, but to point there is no "good" license for fonts, that:
— protects the work (in the libre/GPL sense)
— protects the user, and clearly addresses common font use-cases (partial or complete embedding in other digital documents such as PDF or videos for titling, @font css links Opera made popular)
— has solid attorney-checked wording (seems the OSI or the FSF are the only entities that can do this)
— is understandable by the people producing fonts (the software jargon in the GPL was a plus for software producers but is a minus for everyone else). Pretty diagrams and FAQs are a definite plus
— can be used as-is (in "just copy the authoritative text" or even "I license my font under the fooL" mode), not with bits that need to be hunted down all over the internet or a name that can be confused with something else. Widespread as-is use made the GPL a success.
— proposes a clean way to *optionally* add the "must rename" and "can only be distributed as part of something else" font authors like and Vera experimented (ideally, without polluting the core licensing text file itself with reserved names)

Because it's fine to get excited about Tivos, but there are several ordrers of magnitude between the number of people who use Tivos and the number of people who need libre fonts.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 19, 2009 4:15 UTC (Thu) by vonbrand (guest, #4458) [Link]

"GPLv2 or later" is certainly a problem. The GPL is controlled by FSF, and I won't necessarily agree with some "later" version of the license they could come up with. And nothing guarantees that "GPLv2 only" will stay compatible with later versions either.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 20, 2009 9:01 UTC (Fri) by jpetso (guest, #36230) [Link]

Personally, I think the license incompatibility that comes with
"GPLv{x} only" is much worse than whatever the FSF could come up with.
Once a project has closed down the "or later" option, it's virtually
impossible to link to future GPLv{x+1} software even if it would totally
make sense for the (then) developers from a community and code-sharing
point of view and all.

The FSF changes is only ever going to change stuff that doesn't
significantly affect the free software ecosystem, while doing a
"GPLv{x} only" has a real (negative) impact on possible collaboration
between projects. I find it irresponsible to do that to one's own software.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 19:53 UTC (Wed) by BrucePerens (guest, #2510) [Link]

Regarding the MGOpen license, it requires that you rename derivatives. That's not unreasonable. I agree the text about the current maintainer is silly, but it doesn't give the current maintainer more rights than the copyright holder has. Every license has an asymmetry about the copyright holder.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 8:16 UTC (Wed) by njs (subscriber, #40338) [Link]

...So basically they went to OSI, OSI told them they screwed up majorly, they said "whatever!" and wrote a misleading webpage talking about how OSI-awesome they were, with broad community consensus etc.?

I guess I had pretty mixed feelings about SIL to start with, but that's Double Plus Uncool.

I dunno if the FSF is capable/prepared to write a good font license, but maybe they should try...

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 8:25 UTC (Wed) by BrucePerens (guest, #2510) [Link]

You can go to the OSI site and look at the license-review mailing list archive. All of the correspondence is there. The usual license reviewers, including me, panned the license and Nicholas didn't understand. OSI doesn't say much, they just vote. Sometimes their answer is not voting. That seems to be the case this time.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 10:57 UTC (Wed) by hppnq (guest, #14462) [Link]

There is a loophole in the license that allows you to convert any font under it to public domain.

On the FAQ webpage it says that you cannot remove the copyright, and also the FSF does not seem to find anything funny about this license, which they classify as a "free copyleft license for fonts". Also note this 2006 discussion at LWN, where a similar strong-voiced but not so convincing case was made that this license should not be used.

At the same time, of course, Linux distributions such as Debian include fonts distributed with this license.

So what exactly is this loophole? Why isn't it fixed? Why exactly would this fall apart in a courtroom?

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 16:52 UTC (Wed) by BrucePerens (guest, #2510) [Link]

On the FAQ webpage it says that you cannot remove the copyright
Unfortunately, a statement in the FAQ doesn't effect how a court would interpret the license. The license says this:
The requirement for fonts to remain under this license does not apply to any document created using the Font Software.
The effect of the above is that the license is entirely removed when the font is embedded in a document, and you get the choice of any license you wish. If you then extract the font from the document, you still have whatever license you used on the document.

Sensible license creators, on being informed that there is text of dubious interpretation, would repair the text lest it be read in a way unintended in a court room. The SIL folks, however, only respond with "the developers wanted it that way".

Bruce

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 21:40 UTC (Wed) by hppnq (guest, #14462) [Link]

The requirement for fonts to remain under this license does not apply to any document created using the Font Software.

The effect of the above is that the license is entirely removed when the font is embedded in a document, and you get the choice of any license you wish. If you then extract the font from the document, you still have whatever license you used on the document.

Your train of thought is a bit like that of the guy they found with the smoking gun, claiming that he hadn't fired it and that the smoke was not from the gun.

Say you extract the Font Software from the document, you are left with ... the Font Software, including the license it came with. If there were no license, you would not have been allowed to use the font in your document in the first place. It seems to me that now, you cannot prove that the license you extracted with the software is no longer applicable to the Font Software, because it was once embedded in a document.

It is a wonderful license! I would not change it either. But I am not a lawyer, I would be interested to hear Eben Moglen's take on this.

The intent here is obviously to avoid any uncertainty about the licensing of documents that use OFL licensed fonts, like there is a clause in the Python License about programs written in Python. Given the nature of documents and fonts, obviously something must be said about this use of the font, and frankly, I can't see how they could have made it much clearer.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 21:58 UTC (Wed) by BrucePerens (guest, #2510) [Link]

It's not that there is no license, it's that the license permits the transition to whatever license you wish to apply to a document, when you embed the font in the document. And there's nothing in the license that makes it go back to the original license again.

If the situation isn't fixed, eventually I will just have to publish a site of public-domain versions of SIL-licensed fonts. Not because I don't like the developers, but just to warn others away from the license.

Bruce

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 19, 2009 10:09 UTC (Thu) by hppnq (guest, #14462) [Link]

What I was observing, is that indeed, it is possible to embed the Font Software in a document, put the document under license X, and then extract the Font Software again -- but that it would get you nowhere.

Note that the only case the developers seem to want to protect themselves, is the distribution of copies that are hardly modified, to make a quick buck. So we are only looking at cases in which the extracted Font Software looks remarkably similar to the original work.

Now the question is: how would you point out to a judge that your relicensed Font Software was once embedded in a document? You couldn't. The judge would, I hope, have to conclude that the non-applicability of the OFL is only ... applicable, when the Font Software is actually embedded in a document.

Don't get me wrong: I do understand that there is an entire legal industry built on these simple matters, and I am just toying with these thoughts. But can I ask you why you don't ask a proper attorney to look at this matter, instead of making a blacklist?!

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 19, 2009 16:16 UTC (Thu) by BrucePerens (guest, #2510) [Link]

how would you point out to a judge that your relicensed Font Software was once embedded in a document?

It's very simple to document a process like this for legal purposes. I would create a shell script that carries out thed entire process: embeds font in document, extracts font again. I would place this script online with the intermediate components of the process: the original font, the document before the font is embedded, the document with embedded font, the extracted font.

Processes are presented to courts in many cases.

Bruce

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 19, 2009 17:12 UTC (Thu) by hppnq (guest, #14462) [Link]

Maybe it's good to quote the entire requirement:

5) The Font Software, modified or unmodified, in part or in whole, must be distributed entirely under this license, and must not be distributed under any other license. The requirement for fonts to remain under this license does not apply to any document created using the Font Software.

The OFL clearly states that it is only the document to which the license requirement is not applicable, and the only uncertainty is the precise relation between the document and the fonts used in the creation of that document.

But thanks to your shell script any judge will now be able to clearly separate the Font Software and the document, so I rest my case. ;-)

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 19, 2009 17:34 UTC (Thu) by BrucePerens (guest, #2510) [Link]

Only the one sentence that releases the license matters, because the rest of the license goes away, permanently, as a result of that sentence the moment the font is combined with a document. Nothing in the license can ever govern that copy of the font again, or descendants of that copy of the font, from that moment on.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 19, 2009 20:11 UTC (Thu) by hppnq (guest, #14462) [Link]

the rest of the license goes away, permanently, as a result of that sentence the moment the font is combined with a document

That's not how I read it, but I am not a lawyer.

So let's examine your favourite example, that exposes the truck-sized loophole: embedding a font by dropping it wholely in an OpenDocument zip. It seems trivial to do.

As far as I can see there are three possibilities:

1) The font remains embedded in the document, modified or unmodified
2) The font is extracted unmodified, i.e., including the original license
3) The font is extracted and modified, in particular, without the license

The first case is not a problem. This is what fonts are for.

In the second case, you claim you can now relicense the Font Software, because you have just extracted it from a document. The license you are about to remove clearly refers to the document that you are still holding in your right hand, Font Software in the left. I think the judge will not accept this cheap conjury.

The last case is the trickiest one. But unlike in this example, real-world examples would seem to require a non-trivial effort on the part of the abusing distributor. This is all the developers wanted.

Whether or not it will stand up in court, like all legal matters, will ultimately only be decided in court, and not on websites or in licenses.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 19, 2009 20:27 UTC (Thu) by BrucePerens (guest, #2510) [Link]

2) The font is extracted unmodified, i.e., including the original license
It's unmodified, but the original license isn't there any longer. Sorry, this is a result of what the license says.

Now, if we brought this to court, you would be arguing from what you intended your license to say, and I would be arguing from what your license actually says. If you want to lose a case, having "do what I mean, not what I say" as an argument is an almost sure way to do it.

It's unfortunate that the license is written in a way that a cheap conjuring act can disable it, but that's the case. Changing one sentence will fix it. The SIL group needs to get proper legal counsel and make that change.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 20, 2009 9:17 UTC (Fri) by hppnq (guest, #14462) [Link]

Now, if we brought this to court, you would be arguing from what you intended your license to say, and I would be arguing from what your license actually says.

No. I would point to the clause in the Font Software license you are holding in your left hand, that says it does not apply to the document you are holding in your right hand. I would be arguing that the Font Software and the document are clearly two separate things, and that the non-applicability of the single requirement pertains to the document only, as the license says.

If you would really push me, I would claim that the Font Software had already been embedded in another document, that was licensed under the OFL. I would borrow your script to show how this works.

I would not win immediately. We would have a lot of fun discussing "document", "embedding" and "extraction" and other important terms in court. Which is fine, that's what it's for.

If you want to lose a case, having "do what I mean, not what I say" as an argument is an almost sure way to do it.

Okay, thanks for the tip. ;-)

Changing one sentence will fix it. The SIL group needs to get proper legal counsel and make that change.

Your suggestion on the OSI approval mailing list to change this sentence was already shot down informally by a lawyer who seems to think that this matter should not be fixed in a license. What I have been wondering about, is why you don't ask a proper lawyer to take a look at this (since you seem to care a lot), instead of starting a crusade against OFL.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 20, 2009 9:26 UTC (Fri) by BrucePerens (guest, #2510) [Link]

I will check with Larry, but I'm pretty sure that Larry means tha exact opposite of what you think. Larry is saying that the document could never be a derivative work of the font. He is not saying that the sentence in question does not give away all of your license rights forever.

By the way, I don't know who you are. If you'd ping me via email, I will connect you with Larry directly.

Bruce

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 20, 2009 11:24 UTC (Fri) by hppnq (guest, #14462) [Link]

I will check with Larry

Great, thanks. I appreciate that.

He is not saying that the sentence in question does not give away all of your license rights forever.

I love the double negation! Maybe you could ask him to take a look at the OFL, instead of having him to talk to me? ;-) Thanks for the offer though.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 23, 2009 10:16 UTC (Mon) by nim-nim (subscriber, #34454) [Link]

> Larry is saying that the document could never be a derivative work of the
> font.

I'm quite surprised someone like Larry Rosen is making such a confusion.

Of course the original *text* used in the document is a separate work that could be licensed in many different ways. However the digital *document* (as in text + font + rules making sure one is used with the other) is clearly a derivative of the fonts. It may be not obvious for a corporate document using Arial but you have only to go on a typographer web site where he exhibits his compositions and the way he managed to balance a particular text, font and illustration to realise there is creativity involved. (and yes the USA may have specific font exemptions to avoid copyright requirements in that case and no that does not help at all font users in other countries).

In fact font embedding in digital documents was added precisely to preserve this kind of artistic composition. People who do not care about it just do not use embedding.

And licenses like the GPL are pretty unambiguous on the way GPL material could be used in larger works.

So specific clauses are required to shield users for the embedding case (not to mention professional foundries are very careful not to let anyone forget embedding requires special authorization). And they need to be written by a lawyer. I doubt you can just say "this is not a derivative" when it obviously is from the law POW.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 24, 2009 9:50 UTC (Tue) by hppnq (guest, #14462) [Link]

I'm quite surprised someone like Larry Rosen is making such a confusion.

Ah, maybe he isn't: he is not not saying what you are saying. ;-)

Follow the link given above, especially this mail addresses your points.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 24, 2009 10:15 UTC (Tue) by nim-nim (subscriber, #34454) [Link]

Did you have that in mind ?

« Software license may prohibit certain combinations of copyrightable work and copyrightable font--but not if they are open source licenses! Only
proprietary licenses can impose such restrictions on combinations affecting
*use* of the works. »

Then I'll point the painfully obvious: digital documents are redistributed too. (even if we forget about the printed special case)

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 24, 2009 11:01 UTC (Tue) by hppnq (guest, #14462) [Link]

Did you have that in mind ?

Not at all. I merely assumed some of the confusion you were speaking of was attributable to not having read the source, so I referred you to it.

What is terribly obvious, is that you cannot create the BlahBlah font and use it to either allow or deny distribution of a digital document of which you are not the copyright holder. This seems more to the point, but I must admit I am not sure what you are trying to say.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 24, 2009 12:13 UTC (Tue) by nim-nim (subscriber, #34454) [Link]

Of course you can deny the distribution of a digital document that embeds your copyrighted font. That's how copyright works.

What you can not deny is the distribution of a document that does not include your work.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 20, 2009 10:33 UTC (Fri) by ekj (guest, #1524) [Link]

That, too, depends on jurisdictions. Some jurisdictions are prone to accept by-the-letter interpretations of texts, even when it's clear to everyone that this was NOT what was intended by the one writing the text.

Other jurisdictions generally consider intent, so that if, for example, I write a document that on the overall balance CLEARLY intended to forbid a certain action, that is not voided, even if I misplace a comma somewhere, unintentionally reversing the meaning of a single sentence somewhere.

Still, I agree there'd be benefit in fixing that.

And what's up with the "cannot be sold alone", but "can be sold with software", does that mean if I want to sell the font, I need to include some software (ANY software) together with it ? Can the "software" be #!/bin/sh; cp fontfiles.wherever /usr/lib/fonts/wherever/

What's the purpose of this ? How does it differ from just outright allowing selling the font ?

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 20, 2009 11:22 UTC (Fri) by nim-nim (subscriber, #34454) [Link]

> And what's up with the "cannot be sold alone", but "can be sold with
> software", does that mean if I want to sell the font, I need to include
> some software (ANY software) together with it ? Can the "software" be
> #!/bin/sh; cp fontfiles.wherever /usr/lib/fonts/wherever/

The great enemy of font creators are the people who skim the internet for font files, put them in a huge directory, and sell the result on cd/dvds.

They're ok with pretty much any other kind of distribution, and this clause is here to allow distributing fonts inside an installation package, while preventing basic leaching.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 19, 2009 2:10 UTC (Thu) by i3839 (guest, #31386) [Link]

The wording is probably almost correct, but the problem is that it says that it doesn't have to remain under that license, effectively stripping it away when the fonts are used in a document. And once it gone it doesn't matter what it said or what its original intention was, because it doesn't apply anymore. You follow both the intent and letter of the license when you embed the font in a document. After that, you're free to do what you want.

If it had said that requirements a, b, c, ... do not apply for documents created using the Font Software, then it would have worked as intended, as the license doesn't disappear, but is mostly dormant when the fonts are embedded in a document.

IMHO this is common sense and you don't have to be an attorny to understand problems like this. That said, to word a license in such way that loopholes like this are avoided and still achieves what you want is more tricky, and attornies are probably highly trained to find legal loopholes in anything.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 19, 2009 2:22 UTC (Thu) by BrucePerens (guest, #2510) [Link]

The concept they needed there and did not have was "derivative work". It might have been sufficient to say "A document embedding this font shall not be considered a derivative work of the font."

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 19, 2009 6:02 UTC (Thu) by i3839 (guest, #31386) [Link]

Umm, not sure about that. I've read (Dutch) copyright laws and Berne convention, but couldn't find the term "derivative work" or anything similar. I think it's the much simpler spelling out of "copyright": The right to copy something, in this case fonts. The whole "derivate work" thing comes only into play when you have a license than wants to apply to the whole work. But as far as legal stuff goes I'm a nitwit.

That said, you're probably right that saying that the rest of the document doesn't fall under the font license would have been sufficient (or more simply, that the font license only applies to fonts and nothing else).

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 19, 2009 16:22 UTC (Thu) by BrucePerens (guest, #2510) [Link]

In the Berne convention, it is Article 12, and is called called "Right of Adaptation".

Sorry, I don't read Dutch. Is there an English version of Dutch copyright law?

It's such a fundamental part of copyright that I'm sure it's there.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 20, 2009 1:38 UTC (Fri) by i3839 (guest, #31386) [Link]

I think it's mostly the same as the Berne convention, or at least compatible. A friend of me was a law student, so he had the books. Wouldn't have a clue where it could be found online, or whether there's an English translation of it.

Article 12 in full:

> Article 12
> Right of Adaptation, Arrangement and Other Alteration
>
> Authors of literary or artistic works shall enjoy the exclusive right of
> authorizing adaptations, arrangements and other alterations of their works.

This seems rather obvious, otherwise copyright would only apply to totally unmodified work, which is rarely the case.

So it seems I misunderstood what is meant with "derivative work", I thought it meant a work as a whole which contains (altered parts of) another work, not merely the altered parts themselves. Sorry for the confusion.

All in all it's quite fuzzy what they mean with "a work" anyway. For instance, how I read the stuff there's nothing preventing a program from using GPL libraries while it's not GPL compatible itself, because it are separate works, except if you consider loading the library as some form of copying, which doesn't seem to be the case (IIRC, there's an exception saying that you're always allowed to use a copyrighted work in its intended way: E.g. copying the CD content to PC and push it through the speakers, for programs to run them, etc). The GPL demands that such programs should fall under the GPL as well, but the GPL doesn't apply if the lib isn't distributed together with the program.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 20, 2009 1:48 UTC (Fri) by BrucePerens (guest, #2510) [Link]

The GPL demands that such programs should fall under the GPL as well, but the GPL doesn't apply if the lib isn't distributed together with the program.
Some people think so, but some people think otherwise and it's not been tested in court. So, don't count on it being true. I tell my customers to honor the GPL across dynamic-linking boundaries, for their own safety.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 20, 2009 2:56 UTC (Fri) by i3839 (guest, #31386) [Link]

Caution should be applied of course, but that works both ways. I'm looking at it from the GPLed lib's point of view, and I don't think I can count on the GPL (or any license) to prevent other programs to not use my lib.

Of course this is more of a theoretical problem, as I probably wouldn't choose the GPL for a library anyway, but I'm curious how people can think that it isn't allowed. As far as I can tell using a lib is not enough to be derived work, as you merely follow an interface, and hence copyright won't disallow things like that. Can you explain why some people think that the GPL can apply anyway?

It's not so clear...

Posted Feb 20, 2009 15:24 UTC (Fri) by khim (subscriber, #9252) [Link]

As far as I can tell using a lib is not enough to be derived work, as you merely follow an interface, and hence copyright won't disallow things like that.

Actually it might - it depends on the interface. But what GPL does prevent is distribution of GPLed library and proprietary program as single package. The whole package is clearly derived work of GPLed library (it includes such library, after all) and so GPL applies. You can try to circumvent this by separate distribution of GPLed library and your program - but court may very well decide that the sole purpose of such a distribution was GPL circumvention.

Thus basically the only sane uncovered possibility are "nVidia drivers": when GPLed part is distributed by one entity and proprietary part - by another, unrelated, one. And this is not particularry interesting case for 99% libraries out there...

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 22, 2009 0:52 UTC (Sun) by njs (subscriber, #40338) [Link]

> As far as I can tell using a lib is not enough to be derived work, as you merely follow an interface

Well, that's the question -- is the interface itself a creative work? When you write code that against the interface, does your code end up incorporating creative ideas expressed in that interface? It probably depends on all sorts of awful details in any given case.

Remember that something can be a derived work even if it incorporates no verbatim parts of the original -- look at character copyright, for instance! The rules for determining derivation are complex and poorly-defined (see [1] for a taste!)

So the ability of the GPL reach from a library to an apps using that library isn't ironclad, but there's a reason that you don't see people testing it in court, either...

[1] http://www.ladas.com/Patents/Computer/SoftwareAndCopyrigh...

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 24, 2009 3:30 UTC (Tue) by i3839 (guest, #31386) [Link]

I'm a technical person, and from my point of view, it looks very simple. For a C API, it's nothing else than a list of function names and their parameter lists, with a description of the data structures used and what the functions do. This is all factual information which isn't (or shouldn't be) copyrightable. More complicated header files with macros and inline functions may be copyrightable, but we're talking about the interface here, not header files.

Of course you can elevate it to an insane level and e.g. consider all output of a program potentially derivative work, or lose yourself in too high abstractions, in which cases it quickly becomes an impossible and unworkable situation.

No offence, but any system where you need millions and years to even solve a plain simple cases like SCO versus IBM is IMHO totally broken. In such a system no one wants to go to court for anything at all.

E.g. character copyright seems quite clear to me, a character is just something virtual, which makes it weird in a way, but it's still a creation, even if it's hard to pinpoint where it exist.

The non-verbatim part of programs is more the internal organisation, modularisation, interaction and communication, which is indeed vague for copyright. But merely using one library for its intended purpose is no way near that vagueness. It's a bit more vague when that library wasn't intended to be stand-alone and ripped out of an existing program. But stand-alone libraries are crystal clear IMHO.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 24, 2009 5:45 UTC (Tue) by njs (subscriber, #40338) [Link]

I'm not sure what you're arguing now -- it sounds like you're talking about how things would work if you were in charge ("I'm a technical person", "you can...consider", "any system where ... is broken", "crystal clear IMHO"), which is fine, but I thought we were talking about the legal definition of derivative work, not our personal definitions. All I was arguing was that the legal definition of derivative work is, in fact, less clear than one might expect or prefer.

Also, umm... "internal organisation, modularisation, interaction and communication" -- aren't those *exactly* what a library interface is *about*?

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 24, 2009 21:31 UTC (Tue) by dlang (guest, #313) [Link]

there are explicit exceptions in copyright for interoperability. this is why lexmark lost it's lawsuit against clone printer cartridge manufacturers, even though they copies portions of code bit-for-bit those bits needed to be exactly the same for the cartridge to work and so were part of the interface

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 25, 2009 0:43 UTC (Wed) by i3839 (guest, #31386) [Link]

I'm describing how things work according to my interpretation of all that legal stuff.

> aren't those *exactly* what a library interface is *about*?

Only for internally used libraries. For stand-alone libraries it's their interface to other, independent programs, and it's a clear boundary between one work and others.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 20, 2009 9:25 UTC (Fri) by hppnq (guest, #14462) [Link]

there's nothing preventing a program from using GPL libraries while it's not GPL compatible itself, because it are separate works, except if you consider loading the library as some form of copying

You can link GPL software to any software you like, and copy it a million times. But you may not be able to distribute the result of your work.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 20, 2009 10:05 UTC (Fri) by i3839 (guest, #31386) [Link]

Of course. You can also take pictures of the Mona Lisa, print them out and draw as many moustaches as you want. That's not the point.

What I mean is, as long as you don't distribute the GPL library together with something not GPL which uses it, but only the latter, the GPL can't prevent that because it's out of the picture, as far as I can see.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 20, 2009 11:14 UTC (Fri) by hppnq (guest, #14462) [Link]

as long as you don't distribute the GPL library together with something not GPL which uses it, but only the latter, the GPL can't prevent that because it's out of the picture

No, of course not. And the GPL does not try to prevent this from happening at all, of course.

Even if you're not a lawyer, it is not difficult to at least get a basic grasp of what the GPL tries to prevent, and what not. Read it!

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 20, 2009 22:54 UTC (Fri) by i3839 (guest, #31386) [Link]

I agree with you, and think it's obvious as well, but some other people don't seem to see it that way. I'm curious why that is so.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 7:14 UTC (Wed) by nim-nim (subscriber, #34454) [Link]

The OFL is indeed one of the best choices we have right now, but it's not totally satisfying:

1. it avoided the problem of having to explain to font designers what a “preferred source” is by not making any specific source requirements, so it's really easy to get fonts releases without the bits the original author used to actually create them (and it also confuses authors who understand the need for “preferred source” bu do not see how they fit in the OFL world). Sure you can always edit the result in a GUI font editor, but it's not always as easy as using the original sources.

2. It has provisions for font renaming which I suppose are necessary in some cases but are non-obvious for authors that do not really want to force a renaming of the derivatives (that often rename just in case because the OFL seems to say they must)

So it's neither the BSD nor the GPL of the font world, something in-between not all people like.

OTOH it does not have the font embedding problems plain GPL (as used by too many font projects) has. You can do a lot worse than using the OFL today. It's just, a better license in the GPL class would be nice.

Anyway, none of the aforementioned high-profile font releases used the OFL (Stix may eventually though)

Warning regarding the SIL Open Font License

Posted Feb 18, 2009 8:20 UTC (Wed) by BrucePerens (guest, #2510) [Link]

Here is a copy of the warning I've published regarding the OFL:

The SIL Open Font license purports to be an Open Source license for font developers. In my opinion, font developers would be well advised to steer clear of it.

The SIL guys don't mean to harm anyone, they are just incredibly over-confident in their ability to write an effective software license without legal assistance, and they don't understand what can go wrong when people do that.

The fundamental problem with the license is that its developers are not attorneys and do not appear to have made use of an attorney at any point in the process. They use legal language incorrectly. There appears to be an unintended loophole that would allow the conversion of any font under the license to public domain.

The only power of a license is that a judge will interpret its terms and enforce them in court as part of a lawsuit. For this to work, the legal language has to be written in the way that the judge understands. Attorneys go through long training to do this. Neophytes do not generally attach the same meaning to legal words and are likely to create licenses that don't parse correctly and don't mean what they are expected to.

This would not be so bad if the SIL folks were only using the license for their own work, rather than promoting it to others. An official-looking web site for the license, with text about a multi-year review, and illustrations regarding how the license is expected to work, is likely to lull the unsuspecting developer into trusting the license. That developer is likely to be hurt when the license doesn't perform as expected.

The SIL representative I communicated with claimed that SFLC reviewed the license, but it appears that the license was actually submitted to FSF, not SFLC. FSF's only interest was to determine whether the license was free or not. As far as I can determine, SFLC did not perform a legal review.

Before a version of the SIL license is promoted to users, it should go through a rewrite by a qualified attorney.

Bruce Perens

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 17, 2009 8:28 UTC (Tue) by janpla (guest, #11093) [Link]

I am not entirely sure that a large number of open source licenses is a problem as such - perhaps there is a genuine need; the nice thing about OSS is that we're no longer bound to the "one-size-fits-all" way of thinking.

But what I would like to see is a table comparing them all on their essential features, so we could easily see what they were all about.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 17, 2009 15:24 UTC (Tue) by MKesper (guest, #38539) [Link]

Every new Free Software license is one too much as it creates confusion and has a big chance to be incompatible to existing licenses.
It's like standards: We don't gain anything by inventing new "standards" that are incompatible to existing ones (think MS OOXML vs. ODF).
We may gain by using different approaches to solve problems competing with each other (think KDE vs Gnome vs. ...) and the possibility to learn from each other.

Bruce Perens: How Many Open Source Licenses Do You Need? (IT Management)

Posted Feb 18, 2009 6:41 UTC (Wed) by njs (subscriber, #40338) [Link]

Exercising our freedom to create requires not only the freedom to build on individual pieces of prior work, but also the freedom to combine them.

Therefore, proliferation of free licenses hurts freedom.


Copyright © 2009, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds