All Noto fonts now licensed under Open Font License 1.1

[Doug Felt]

With the new release we have moved from Apache 2.0 to Open Font License 1.1 (OFL 1.1).  We are making this change because it makes it easier for you, the end users of Noto, to use the fonts that we produce.  When we first released the Noto fonts the OFL license was still a bit of an unknown entity and the Apache license looked familiar to us.  Over the years the OFL license has become the most commonly used and understood license to use for open source fonts.

All of our fonts going forward from this release will be available only under the OFL license.  If you still want the fonts under the Apache license you can get them from the repo using the tag 'v2015-09-29-license-apache'.  However, new fonts and updates to existing fonts will use the OFL license.

(on behalf of the noto font team)

[Dave Crossland]

Thanks Doug! This is fantastic news! Congratulations to everyone on the Noto team for making this happen :) 

[Adam Twardoch]

Erm, right. :/ 

[Behdad Esfahbod]

On 15-11-27 02:34 PM, Adam Twardoch wrote:
> Erm, right. :/

Care to elaborate?

> On Wednesday, September 30, 2015 at 2:47:18 AM UTC+2, Dave Crossland wrote:
>
> Thanks Doug! This is fantastic news! Congratulations to everyone on the
> Noto team for making this happen :)
>

> --
> You received this message because you are subscribed to the Google Groups
> "noto-font" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to noto-font+...@googlegroups.com
> <mailto:noto-font+...@googlegroups.com>.
> To post to this group, send email to noto...@googlegroups.com
> <mailto:noto...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/noto-font/b6caa8a7-c4d9-42cc-9832-defde24f6688%40googlegroups.com
> <https://groups.google.com/d/msgid/noto-font/b6caa8a7-c4d9-42cc-9832-defde24f6688%40googlegroups.com?utm_medium=email&utm_source=footer>.
> For more options, visit https://groups.google.com/d/optout.

[Adam Twardoch (List)]

The great thing about the Noto fonts being available under the Apache 2 license was that it made them far more useful than the mere fonts by themselves.

I consider making the Noto fonts available to the world under the permissive Apache 2 license a very significant contribution to the world literacy, because, in addition to them being just "more fonts", they can also serve as permissively-licensed repositories for what Thomas Milo calls "script grammars".

The selection of the glyph repertoires relevant for each given script, to some extent the glyph naming, the structure of which glyphs are used as components in composite glyphs, the positioning of the components, the mark attachment anchors, the GSUB OpenType Layout features that describe orthotypographic relations between glyphs that go beyond and greatly enhance the info encoded in the Unicode standard — all that is an excellent reference and research material, available in the best possible way, i.e. working fonts that can be dissected, analyzed, and portions of which can be reused in other projects.

This can go from simple reusal of GSUB tables or their subsets in form of "enhanced font templates", which can potentially lead to creation of more fonts, especially for complex scripts. Some of those 3rd-party fonts could be released as opensource as well, but that would need some funding, or the work will be done by amateurs as hobbyists (as the case was before). But, with Apache 2-licensed "components" (which the Noto fonts are excellent for), the knowledge how to make functioning fonts for all scripts in the world is finally far more attainable than ever before. So even if the Noto components were used to create commercial fonts, that would still enhance the typographic market in different regions, contribute to more visual diversity, higher cultural awareness. To me, the Apache 2-licensed Noto fonts are a "toolkit" for democratization of type outside the Western part of the world, similar to how Fontographer was just that for the West 25-30 years ago.

The Apache 2 license is not the most liberal (permissive) in the world, but it certainly is more permissive than SIL OFL. The wording of the OFL, and its intended interpretation, suggests that I cannot take any part of an OFL font (say the GSUB table) and put them into a commercial font (I could explain this in detail, but you may choose to trust me on this). Given this restriction, the potential "network effect" of the Noto font project as a significant contribution to the world’s written culture is severely limited.

Of course, with the Noto fonts being Apache 2, it makes it much more difficult for you guys to source from the ever-growing OFL repository and *add to* the Noto fonts from other OFL fonts. But with Noto being Apache 2, it limits (in my view) the usefulness of Noto as a permissive source of script-related typographic information that goes beyond "just fonts". Essentially, it makes the Noto fonts "just fonts", it makes them "linguistic and typographic software tools", on par with fontTools, HarfBuzz and Unicode.

Don't get me wrong — I still think that the fact that the Noto project exists, is actively maintained, and constantly improves. Exactly because of that, there is (or had been) an implicit guarantee that the script-related and typographic information "encoded" in the Noto fonts would be increasingly correct, and, over time, would increasingly accurately reflect the wide public’s view of what’s relevant or good for reading. Which made it a great resource — in my opinion even better than the Unicode standard, which to some extent is a bit “academic”, while the Noto project is much more practicality-driven and “down to Earth”.

To me, in this particular context, the actual glyph images contained in the Noto fonts are slightly less-valuable. They’re still useful for potential reference, and — in case of various mathematical or technical symbols or other generic glyphs — would actually be superbly useful. In fact, I would be the first to try to add weight and width variation into something like Noto Sans Symbols rather than Noto Sans "proper", and trying to use the GX Variation technology to develop smart fallback fonts for the "long tail" of lesser-used Unicode characters or scripts. That effort would be aided by the Apache 2 license as well, because professional designers could be brought to the project who could do this work *if* they "get something out of it", i.e. a "flexible outline generation machine" for their *own* fonts, so they don't have to draw ⌀, ⌘ or ⋠ all over again.

I attribute Wikipedia’s great success to the fact that its contents is very permissively licensed, so people contribute, because they (or others) can also benefit. It’s about sharing knowledge. I have considered the Apache 2-licensed Noto project to have a pivotal role in knowledge sharing within digital typography, with significant long-term benefits that are much larger the goal of the end-user having "at least one glyph for each Unicode character". Switching to OFL limits that potential, though it may have some benefits for end-users in the immediate term.

All the best!
Adam

> To unsubscribe from this group and stop receiving emails from it, send an email to noto-font+...@googlegroups.com.
> To post to this group, send email to noto...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/noto-font/5658B095.4020704%40gmail.com.

[Adam Twardoch (List)]

> On 27 Nov 2015, at 21:46, Adam Twardoch (List) <list...@twardoch.com> wrote:
>
> Essentially, it makes the Noto fonts "just fonts", it makes them "linguistic and typographic software tools", on par with fontTools, HarfBuzz and Unicode.

This should read:

Essentially, OFL makes the Noto fonts "just fonts", while Apache 2 makes them "linguistic and typographic software tools", on par with fontTools, HarfBuzz and Unicode.

A.

[Behdad Esfahbod]

Adam,

I fully agree with you on the benefits of an open Noto ecosystem. Indeed, I
licensed HarfBuzz as Apache, instead of LGPL, exactly because I wanted the
widest adoption possible. However, Apache for fonts is seriously problematic:

On 15-11-27 03:46 PM, Adam Twardoch (List) wrote:
> The Apache 2 license is not the most liberal (permissive) in the world, but it certainly is more permissive than SIL OFL.

That might be true, but Apache is cumbersome. It literally meant that if you
use the font to print something and hang it in a museum, you'd need to hand
the Apache license and Noto copyright next to it! I'm not talking
theoretically; we actually got such enquiry. So we were left to decide
between a few different alternatives, none them truly satisfactory:

1. live with Apache for fonts and keep telling everyone that our intent is
different from our license,

2. add a font clause to Apache or otherwise invent a new license. I don't
think *anyone* likes that,

3. dual-license Apache and OFL, or whatever other two licenses. This has it's
own problems: a) dual-licensing severely limits what contributions you can
take in (this is not a big deal for Noto), but more importantly b) OFL
explicitly disallows dual-licensing. I don't agree with them, but that's the
way it is.

> The wording of the OFL, and its intended interpretation, suggests that I cannot take any part of an OFL font (say the GSUB table) and put them into a commercial font (I could explain this in detail, but you may choose to trust me on this). Given this restriction, the potential "network effect" of the Noto font project as a significant contribution to the world’s written culture is severely limited.

I understand your point. We can (and I think we should), work around this, by
somehow licensing our *source* materials differently. Let me take a look at
the options and come up with a proposal.

Everyone, please discuss!

Cheers,
behdad

[Adam Twardoch]

> On 27 Nov 2015, at 21:57, Behdad Esfahbod <behdad....@gmail.com> wrote:
> It literally meant that if you
> use the font to print something and hang it in a museum, you'd need to hand
> the Apache license and Noto copyright next to it!

Behdad,

I see how this can be problematic. In different jurisdictions fonts are copyrighted in different ways. In the U.S. digital fonts are protected by copyright only as "software programs", but in some countries, digital fonts can be protected akin to "works of art" (for example in the Russian Federation).

The Apache 2 license does not limit its definition of the term "Work" to software programs. It alludes to it, and uses the word "software" explicitly in the Appendix boilerplate, but that's not a constituting part of the license. In other words, in theory people can publish types of work other than software programs (for example music) under the Apache 2 license.

So, while in the U.S. publishing the Noto fonts under the Apache 2 license has to be interpreted in such a way that the license only applies to the Noto fonts "as software programs", in other jurisdictions, it's less clear. For example, some jurisdictions could see it so that the Apache 2 license applies to the Noto fonts as "works of art".

That, of course, influences the understanding of what a Derivative Work is.

If you take the U.S. point of view, that the Work protected by copyright (and therefore, the Work to which the Apache 2 license applies) is "software programs", or more specifically "font software programs", then the Derivative Work would need to be another software program. With software programs, U.S. copyright does not extend its protection to the *output* of the software program. In other words, if I use Adobe Illustrator to draw a graphic, then Adobe does not have copyright claims on the graphic I’ve drawn, because the graphic is an output of the Adobe Illustrator program, and *not* a derivative work of Adobe Illustrator. Similarly, if I use FontLab VI or FontForge to create a font, that font is not a derivative work of the font editor, it's the output of the font editor.

If I use the Noto fonts to create a digital image or to print something that I’ve typeset, that digital or printed image is *not* a Derivative Work of the Noto fonts "software program", because the result is not another software program. The result is *the output of the software program*, and that’s not a derivative work.

However, even in the U.S., a subsetted font embedded in a PDF document can, indeed, be seen as a Derivative Work of the Noto font software program, so the Apache 2 requirement "You must give any other recipients of the Work or Derivative Works a copy of this License; and You must cause any modified files to carry prominent notices stating that You changed the files" may still apply, which is indeed problematic.

In addition, when the Noto fonts are considered "works of art" in some jurisdictions, then indeed a printed image of text typeset in Noto could be interpreted as being a "reproduction of the work of art", and therefore, a Derivative Work under the Apache 2 license. In this case, your "hanging in a museum" example may, indeed, apply.

One immediate remedy to eliminate at least a part of the problem would be to change the wording "Licensed under the Apache License, Version 2.0" in the font's License "name" table field to "This font software is licensed under the Apache License, Version 2.0". In other words, you could specify more clearly the *nature* of the Work that is covered by the license, say that the Work as *you* see it is "font software" rather than something else. I.e. you'd codify the U.S. legal practice. The phrase "font software" is widely used in font EULAs, and is, indeed, use explicitly in the SIL Open Font License.

This path would not, alas, eliminate the problem fully. If a full verbatim copy of the Apache 2 license were placed inside the Noto fonts' "name" table, then any conversion or embedding processes that do not remove that information would satisfy the requirement "You must give any other recipients of the Work or Derivative Works a copy of this License". The requirement "You must cause any modified files to carry prominent notices stating that You changed the files" is still a bit vague, because of course in most cases, the user does not have influence over the exact technical process that happens during font embedding.

For all those reasons, of course, the SIL OFL license is more practical, because it is more font-specific and it deals more closely with those font-specific edge cases, and explicitly exempts "any document created using the Font Software" from the license-change prohibition.

However, the OFL requirement "Original or Modified Versions of the Font Software may be bundled, redistributed and/or sold with any software, provided that each copy contains the above copyright notice and this license" is actually similarly problematic as the Apache 2 requirement to "give any recipient a copy of this License". In that aspect, neither Apache 2 nor OFL really permit the style of subsetting & minification that the fontTools subsetter performs by default, where it zaps most of the "name" table (or does it not?). Either way, both Apache 2 and OFL actually are equally adamant about "giving a copy of the license to the recipient" under some circumstances, and some scrutinous lawyers may be bothered with both these licenses equally.

Really, the only *real* problem with the Apache 2 license is that it requires users to "cause any modified files to carry prominent notices stating that You changed the files". That's a restriction that the OFL does not have.

The "hanging in a museum" problem is equally good, or equally bad, depending on how you see it, in both Apache 2 and OFL. These two licenses are really quite alike.

And, finally, the only *other* problem with Apache 2, one that is easily addressed, is the notion of what exactly, in case of the Noto fonts, is "the Work" that the Apache 2 license covers. Clarifying the scope and context of the term "Work" as it applies to the Noto fonts, via the wording "This font software is licensed under the Apache License, Version 2.0", would, in my opinion, fully suffice to ensure that documents, images, vector graphics, printouts, vinyl signage etc. would *not* be considered Derivative Works, because they'd be output of software rather than another piece of software.

Oh: I'm not a lawyer. But more often than not, lawyers agree with me when it comes to font licensing stuff. ;)

Best,
Adam

[Adam Twardoch]

> On 27 Nov 2015, at 22:44, Adam Twardoch <ad...@twardoch.com> wrote:
>
> And, finally, the only *other* problem with Apache 2, one that is easily addressed, is the notion of what exactly, in case of the Noto fonts, is "the Work" that the Apache 2 license covers. Clarifying the scope and context of the term "Work" as it applies to the Noto fonts, via the wording "This font software is licensed under the Apache License, Version 2.0", would, in my opinion, fully suffice to ensure that documents, images, vector graphics, printouts, vinyl signage etc. would *not* be considered Derivative Works, because they'd be output of software rather than another piece of software.

Ps. Even the more generic wording "This font is licensed under the Apache License, Version 2.0" would suffice. In pretty much all jurisdictions I can think of, font protection does not extend itself to works created using the font. In other words, pretty much all jurisdictions assume that a derivative work of a font can only be a font. An article typeset in a font and printed is *never* really a derivative work of the font that was used to typeset it. This has nothing to do with Apache or not Apache. This is something that has been solved many times over in the last 100 years before fonts were digital, and has been solved after they have.

So I'm only recommending prepending the "Licensed under..." with "This font is..." or "This font software is..." as an extra precaution because Apache 2 very explicitly talks about the Work and the Derivative Work. Any sensible lawyer would not need this extra precaution, but they would have to arrive at the conclusions themselves. So stating it explicitly is helpful.

So, really, one problem with Apache 2 is easily fixed, another does not fully go away with the switch to OFL.

It's only the "must cause any modified files to carry prominent notices stating that You changed the files" clause that is problematic. But — if the Noto fonts were digitally signed (using a DSIG table), then the absence or invalidation of the digital signature (which invariably would happen in any scenario that modifies the font) would in my view* be a sufficient resolution of that problem.

*) Of course some people may disagree with that. But if you look at that Apache 2 clause, it does not require me to put into the notice *who* made the change, just *that* it was changed. The point of this claus is that people don't have to go find the original Work and then they don't have to *compare* it to the copy of the Work they have to find out whether the copy they have is modified or not. In other words, the copy of the Work needs to carry a mechanism that allows anyone to verify whether the Work has been modified or not. And for that, the OpenType DSIG mechanism is just perfectly suited. Many other types of software programs don’t have such a mechanism.

So all in all, a no-brainer. :)

Best,
Adam

[Adam Twardoch]

Ps. I would happily help to author a FAQ that explains clauses of the Apache 2 license in context of how it applies to the Noto fonts.

Yesterday, I went a bit paranoid and hastily made a fork of the Apache 2 Noto fonts myself ( https://github.com/twardoch/toto-fonts/ because if "Noto" went on to mean "no to permissive licensing" then "Toto" should mean the opposite, or something :D ).

So I'll work a bit on an Apache 2 clarification FAQ there myself, and let you know when it's done. Then, you can decide how to proceed.

And — to make things perfectly clear: there is no problem with dual-licensing of Noto *for you*.

The OFL clause "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" does NOT apply to the copyright holder. The copyright holder, i.e. in case of the Noto fonts Google, is *not* bound by the terms of the license under which he chooses to share his work with the world. That could lead to interesting self-referential paradoxes :)

No no, Google as the copyright holder can do whatever they want with the works it owns. Including publishing it under one license, or two, or twenty. It is *only* the license recipients who are bound by the terms of a given license. FreeType is published under FTL and GPL, although the GPL does not permit dual-licensing either. But when software is dual-licensed, recipients can choose the terms under which they want to use or distribute the software. The one license does not have the power to "spill over" to the other one.

The only complication would arise if Google wanted to take take some OFL-licensed glyphs from 3rd party fonts and add them to some Noto fonts. Then, the result would have to be OFL, but that's because Google would be bound by the OFL licensing terms as a licensee, a license *recipient* from the 3rd party. Or if somebody would take an existing OFL-licensed Noto font, extend it, and publish his own fork also under OFL. Then Google would not be able to take ownership of that or publish that under Apache 2 — except of course if you go talk to that person and convince him to also release their modifications under Apache 2. Plus, that stuff does not really happen often.

So dual-licensing under Apache 2 and OFL is very much possible. As long as there are no complicated forks and hijacks, i.e. as long as it's one and the same font file licensed under two different licenses, there is no problem. In case of FreeType, some person could theoretically fork it, and release an extension *only* under GPL, not under FTL. Then, the FreeType authors could not flow these changes back into the mainstream code if they wanted to keep the ability to dual-license.

So Apache 2 + OFL could work for Noto, because any user could choose the more favorable option that suits their needs more fully. As long as you don't plan to import glyphs from OFL-only fonts into Noto, and you plan to fully own copyright over Noto, there is no need to worry. If people fork or create derivatives under only one license, and you don't care about those forks, let them be.

So, all in all, I would recommend:

1. Release of the Noto fonts in a dual-licensed manner, under Apache 2 and OFL (because I see benefits of the OFL simply being more explicit and specific about the permitted usage).

2. In the License field, say:

==
This Font Software is licensed under the Apache License, Version 2.0, and under the SIL Open Font License, Version 1.1. This Font Software is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See either the Apache License or SIL Open Font License for the specific language, permissions and limitations governing your use of this Font Software.
==

3. Possibly (regardless of whether it's Apache or OFL), consider finding language for an extra sentence in the License field that would say something to the effect of "This notice effectively fulfills the requirement of including a copy of the license", or something like that. That would allow you (and the derivative makers, document creators etc.) to skip the requirement to include the full text of the license in the font files or somewhere else.

The "name" table is not really the best place to host these huge license texts. This is something you SHOULD do anyway, because, again, otherwise, being strictly legal, if YOU release the Noto fonts as they are, be it under Apache or OFL, ANY user who makes even a hint of derivation, would actually have to put the FULL OFL or FULL Apache or BOTH somewhere, in some cases inside the font's name table, even though that text isn't there yet. In other words: you as copyright owner don't have to put these texts there, but anybody who even embeds one of your fonts into a PDF, theoretically would have to. Again, regardless of whether it's OFL or Apache :) That's the problem with both these licenses.

The above strategy would also take care of the caveat of the "must cause any modified files to carry prominent notices stating that You changed the files" Apache 2 clause: if somebody is a normal end-user, they can choose the OFL terms and then not be bound by this requirement. If somebody is a "derivative maker" or "researcher" of any kind, then they will most likely want to use some portions of Noto, and then the Apache 2 license would be more attractive to them, and then they obviously would need to put some notice such as "Portions..." something — which already by itself fulfils the requirement.

Best,
Adam

[Adam Twardoch]

> So dual-licensing under Apache 2 and OFL is very much possible. As long as there are no complicated forks and hijacks, i.e. as long as it's one and the same font file licensed under two different licenses, there is no problem. In case of FreeType, some person could theoretically fork it, and release an extension *only* under GPL, not under FTL. Then, the FreeType authors could not flow these changes back into the mainstream code if they wanted to keep the ability to dual-license.

So, Apache 2 + OFL for Noto is just as possible and convenient as FTL + GPL2 for FreeType.

Dual-licensing of Noto only becomes slightly complicated if you want to "import" 3rd party OFL glyphs into Noto and take that upstream without explicit permission of that 3rd party (but you could always ask for permission). And the other complication would be in case of a "hostile" fork, where somebody takes an Apache 2 + OFL licensed font, extends it, and insists on only releasing it under OFL, and you really want to take those changes upstream.

But, again, if that person is not "hostile" and just releases the changes under the same dual license, then you have no problem whatsoever. I don't think such a "hostile" scenario ever happened in case of FreeType, which is dual-licensed, so it's a very very marginal, theoretical scenario.

Continued Apache 2 + OFL dual licensing would certainly make both end-users and people "like me" happy. :)

I'll say these things again, because these are fundamental things about licensing that many people are confused with:

1. A license (such as OFL, but also any other) does *not* apply to the copyright owner. The copyright owner *owns* the work, and is not bound by the license that he himself worded :)

If Łukasz Dziedzic is the copyright owner of Lato, and he released Lato under OFL, then this license is *his* contract with *the outside world*, but that outside world does not include him. Everybody else may only release Lato derivatives under OFL, but he as the copyright owner, can do whatever he wants.

2. When software is dual-licensed, recipients can choose the terms under which they want to use or distribute the software.

In dual-licensing, the two licenses are "OR". You choose which one applies to you, and then only permissions and restrictions of that license apply to you. Of course you can *choose* to have both licenses apply to you. In other words, if a font is licensed under Apache 2 + OFL, then as a user, I can choose to follow only Apache 2, or only OFL, or both at the same time. And if I want to make a derivative work, I can publish that only under Apache 2, or only under OFL, or under both. The latter would be the "civil" option, and anyone who will deliberately "hijack" that will typically by frowned upon by the community.

OK, I think I'm done for today :D

Cheers,
Adam

[Adam Twardoch (List)]

OK, one more thing on dual licensing, and I'll stop elaborating (you asked for it Behdad ;) ):

A pre-made license isn't some magical entity. It's a template of words that you, the copyright owner, say to the world. It is always *your* permissions and *your* restrictions, not SIL's or FSF's or anyone else's.

They just happen to be identical to the permissions and restrictions that somebody else said to the world, hence the recipient of permissions and restrictions from various copyright owners needs to track fewer variations. That's license compatibility.

If OFL prohibited the *copyright owner* from dual-licensing, then Bree Serif, PT Sans or PT Serif would not exist, because they're also dual-licensed: they're released under OFL and under commercial licenses (as "Pro"). No, that's faulty thinking.

With dual licensing, it's a bit like this: your mother tells you to take off your shoes if you're at her house, and your wife permits you to wear shoes when you're at home with her.

Your mother's permissions or restrictions only apply at her house, not when you're at home with your wife. When you were a kid and you only lived at your parents' house, then your life was dictated exclusively by your parents' permissions and restrictions (single licensing).

But if you now spend some time at home with wife and some at your parents's house with mother, you choose the licenses depending on where you choose to be (dual licensing).

The OFL forbids *derivative makers* from making dual-licensed derivations of fonts that the *copyright owner* released *only* under the OFL. *Only* that.

If Google releases Noto only under OFL, then derivative makers cannot release their derivatives under OFL + Apache 2 or OFL + commercial. But Google can still at any later point add another license because Google is the copyright owner.

If Google releases Noto under Apache 2 + OFL, then OFL's restriction only applies *if the derivative maker chooses to follow OFL*.

But the derivative maker chooses to follow Apache 2, or a combination of Apache 2 + OFL, then the dual-licensing restriction does not apply.

That's logical "OR".

If mother forbids me to eat the yoghurt but father explicitly allows me to eat the yoghurt, I can eat the yoghurt. Mother cannot punish me, because father allowed it. I'm legally in the right to eat the yoghurt. That's sort of (not quite) dual licensing :)

A.

[Dave Crossland]

Hi

On 28 November 2015 at 07:12, Adam Twardoch (List) <list...@twardoch.com> wrote:

OK, one more thing on dual licensing, and I'll stop elaborating (you asked for it Behdad ;) ):

Do you want me to take down Adam's reasoning piece by piece? 

I attribute Wikipedia’s great success to the fact that its contents is very permissively licensed 

Wikipedia uses the copyleft Creative Commons license, CC-BY-SA. 

I attribute Wikipedia’s great success to the fact that its contents is copyleft licensed.

We can (and I think we should), work around this, by somehow licensing our *source* materials differently. 

I think this is a terrible idea. 

Simplicity is crucial to wide adoption and the OFL prohibits dual licensing, so I think keeping everything OFL is important. 

The Noto project already has a lack of community engagement from sources because they are not yet provided, and if they are provided under a complex licensing scheme, that will surely add friction to contributions from people at institutions with legal teams who will need to review the custom scheme. Whereas if it is a single license, already used across Adobe, Microsoft, Apple, Monotype, SIL, etc etc, then its much easier for people there to contribute meaningfully.

With Noto under OFL, the fonts and sources still serve as script grammar references for developers of proprietary licensed fonts, but they have to carry their own water. The working fonts and sources can still be dissected and analyzed, its only that portions can not be reused in restricted fonts. 

Those developers have to do their own work, and that seems fair to me. Doing so will in way increase the active knowledge of the script by the development community, by making those developers more 'fit' - by doing their own work, they will grow stronger from the exercise. 

Adam's idea about making a GX machine for ⌀, ⌘ and ⋠ is fine, and he still has the Apache versions to do that. There's no need for further Apache licensing for Adam to do what he wants, there.

I think what Adam is leaving out when saying that "copyleft is antisocial" (or like cancer or whatever)  is that making liberally licensed Apache stuff proprietary is exactly the same thing. It's just the other side of the same coin. It's hypocritical, like if we were kids in the playground, Adam would be the boy who wouldn't share icecream with me, but then when I have my own and his is gone, he's crying for me to share.

Cheers

Dave

[Dave Crossland]

On Nov 28, 2015 5:52 AM, "Adam Twardoch" <ad...@twardoch.com> wrote:
>
>  FreeType is published under FTL and GPL, although the GPL does not permit dual-licensing either.

This isnt true either. GPL requires *compatible* licenses, but OFL says "only this license and no other."

[Dave Crossland]

On Nov 28, 2015 5:52 AM, "Adam Twardoch" <ad...@twardoch.com> wrote:
>

> The "name" table is not really the best place to host these huge license texts. This is something you SHOULD do anyway, because, again, otherwise, being strictly legal, if YOU release the Noto fonts as they are, be it under Apache or OFL, ANY user who makes even a hint of derivation, would actually have to put the FULL OFL or FULL Apache or BOTH somewhere, in some cases inside the font's name table, even though that text isn't there yet. In other words: you as copyright owner don't have to put these texts there, but anybody who even embeds one of your fonts into a PDF, theoretically would have to. Again, regardless of whether it's OFL or Apache :) That's the problem with both these licenses.

Only Apache has that problem. The OFL explicitly permits document embedding without this requirement.

[Dave Crossland]

And that problem is inherent in the Apache license; one could argue that your toto repo is in violation of the Apache license, as are all derivatives that include any piece of Noto, because they must include the full text of the license. You can verify this by going to any Android device, Settings, About, Legal, Open Source. Or Gmail app, Settings, About, Open Source licenses. These are proprietary applications, but Apache is clear the full license text must accompany all derivatives.

[Adam Twardoch]

Dave,

> With Noto under OFL, the fonts and sources still serve as script grammar references for developers of proprietary licensed fonts, but they have to carry their own water. The working fonts and sources can still be dissected and analyzed, its only that portions can not be reused in restricted fonts.

Well, with sources available in a digital format identical to the format that other people will prepare, and these structures being "script grammars" i.e. largely generic information that simply codifies orthotypographic practice, strong similarity in implementations will be unavoidable. It’s just that it would be “nicer” that there won’t be any accusations.

> Doing so will in way increase the active knowledge of the script by the development community, by making those developers more 'fit' - by doing their own work, they will grow stronger from the exercise.

Right. That’s why we all send children alone to the woods for a month instead of sending them to school where they can learn to collaborate. Survival of the fittest, cool.

> I think what Adam is leaving out when saying that "copyleft is antisocial" (or like cancer or whatever) is that making liberally licensed Apache stuff proprietary is exactly the same thing.

I'm not saying it’s antisocial or cancer. It just excludes the possibility of valuable contributions.

I and many of my colleagues work in the commercial font industry, we earn making money making fonts under a large number of different licensing models.

Quite a few of us are willing to contribute our time (often spare unpaid time) into a shared pool of knowledge, fonts or code if we have a possibility to benefit from this shared pool ourselves in our professional endeavors.

Risk sharing lies at the heart of professional collaboration. I'll choose to participate in a permissively licensed open-source project and will contribute some of my exclusive knowledge, skill or other assets to such a shared pool, but it's far more likely to happen if I also see a chance to, at some later point of time, gaining something from that pool — meaning that I can have the safety that I can use this in all types of my work and not just some.

But if the licensing model of the pool is such that I will contribute but *only* otgers will benefit, because the nature of my work prevents me from benefitting — then I’ll pass. Or will demand straight and upfront money.

Copyleft code is just as proprietary as proprietary code. It just belongs to “some” people, who are *not me*. If I cannot use copyleft code because I’m not permitted to, then this surely feels like proprietary.

I don't mind copyleft, just as much as I don't mind stern Christians. I let them do their thing but I won't be part of it because I don't like to be told what I "must" or "must not" do.

So, more specifically — if you ask me to review or contribute to a permissively licensed opensource project, then I’ll do it on my own risk, meaning in my “spare time”, without immediately asking you to pay me for my time. It’s my risk because it means that I may then try to make some money off it later, with a different client, or not, but you don’t have to worry about it. You benefit from it now, and I may or may not benefit from it later.

But if the project is not permissively licensed, then the only condition I would contribute would be of you pay me now. Because I know that I won’t be able to use it later, or it’ll be complicated and limiting.

As long as you keep coming ’em $$$ my way, I may do work on some OFL or GPL stuff. But if it’s Apache 2 or BSD or MIT, I’ll take a much bigger risk and will work “for free”, hoping or calculating that the money comes from other sources.

I’m using “me” just as an example. FontShop’s FontQA was GPL and nobody touched a single line of it. fontTools & the UFO stuff is permissive, and see how alive it is, and how many people from the commercial world are active in it. Why? Because they know that they’re not wasting their time working on it.

They may not quite have figured out what their own great thing will be where they’ll use it but at least they know now that if their great new idea hits them, they won’t be out of options because they’ve assigned their code copyright to their employer, or put it into a pool from which they can benefit only in an extremely limited way.

If I look at Apache 2-licensed Noto, I see certain limitations, but I also see tremendous potentials, ideas just lit up in my head. If I look at OFL-licensed Noto, I primarily see all those “don'ts”. I instantly know that I must obey. It's like the New vs. Old Testament, basically.

So, well, I’ve had some very cool plans and ideas but if Apache 2-based Noto is a dead end (frozen, no updates coming), then, well, I’ll just scrap those ideas altogether. I have other things to do.

But hey, please keep taking down my reasoning ;)

A.

[Adam Twardoch]

On 28.11.2015, at 02:27, Dave Crossland <da...@lab6.com> wrote:

>  FreeType is published under FTL and GPL, although the GPL does not permit dual-licensing either.

This isnt true either. GPL requires *compatible* licenses, but OFL says "only this license and no other."

Ok, that was an oversimplification. GPL3 allows "sublicensing" under compatible licenses. But that has nothing to do with owner's dual licensing. 

http://www.freetype.org/license.html :

==

FreeType comes with two licenses from which you can choose the one which fits your needs best.

• The FreeType License (FTL) is the most commonly used one. It is a BSD-style license with a credit clause and thus compatible with the GNU Public License (GPL) version 3, but not with the GPL version 2.

• The GNU General Public License (GPL), version 2. Use it for all projects which use the GPLv2 also, or which need a license compatible to the GPLv2.

==

So, FTL is compatible with GPL3 and NOT compatible with GPL2. 

If FreeType wanted to be compatible with just GPL3, they wouldn't need to dual-license. They'd single-license under FTL, period. Compatible licensing and dual- or multi-licensing are two different things. Please don't throw them in here as they have nothing to do with Noto :)

FreeType is dual-licensed under FTL (which happens to be compatible with GPL3) and GPL2 (because FTL is not compatible with it, which is exactly why FreeType is dual-licensed). 

Dual-licdnsing means: you as copyright owner issue two sets of permissions and restrictions, and the user picks the one that suits him better and is then free to completely disregard the other. Even if the other invokes the devil incarnate. 

Apache 2 + OFL is like the world where you can be atheist or Catholic. If you choose to be atheist, then God, Hell, Satan and sin simply do not apply to you. 

A.

[Adam Twardoch]

On 28.11.2015, at 02:39, Dave Crossland <da...@lab6.com> wrote:

> Only Apache has that problem. The OFL explicitly permits document embedding without this requirement.

OFL says: 

==

Permission is hereby granted, free of charge, to any person obtaining a copy of the Font Software, to use, study, copy, merge, embed, modify,

redistribute, and sell modified and unmodified copies of the Font Software, subject to the following conditions:

(...)

2) Original or Modified Versions of the Font Software may be bundled, redistributed and/or sold with any software, provided that each copy contains the above copyright notice and this license. These can be included either as stand-alone text files, human-readable headers or in the appropriate machine-readable metadata fields within text or binary files as long as those fields can be easily viewed by the user.

(...)

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.

==

OFL only explicitly mentions "documents" in condition 5, where it clarifies the "output of software program vs. derivative software program" issue. 

It says that while the fonts (including fonts embedded in documents) are subject to the OFL, the documents created using the fonts are not subject to the OFL. That is also true with Apache 2-licensed fonts, though is not stated explicitly. 

Fonts embedded in documents *are always* subject to OFL, and no provision of the OFL lifts that. Embedding is listed along  with other permissions list, and is "subject to conditions".   

Condition 2 says “bundled, redistributed and/or sold with any software” and not ”bundled with, redistributed with and/or sold with any software”. Which means that the enumeration must be seen separately, i.e. the condition applies to when you either "bundle the fonts" (in any way), or when you "redistribute the fonts" (in any way, including by themselves), or when you "sell the fonts with any software". 

The "with any software" bit applies only to "selling", not to all three. Otherwise it would be ridiculous: the entire condition 2 would only apply to situations if the fonts appear in the "with any software" context. 

So I were to redistribute the fonts by themselves, *without* any software, I wouldn't have to include the copyright notice or the license at all. That would be crazy. 

So condition 2 in fact applies to, anong others, ALL forms of *redistribution* of the fonts, which must obviously include document embedding. This means that embedded fonts are subject to condition 2, i.e. they must "contain this license".

Condition 5 does *not* lift this (or if it does, please show me where). The sentence "The requirement for fonts to

remain under this license does not apply to any document created using the Font Software" *only* addresses license *change*, not the other conditions. 

Down piece by piece it goes. :D 

A.

[Adam Twardoch (List)]

On 28.11.2015, at 02:39, Dave Crossland <da...@lab6.com> wrote:

These are proprietary applications, but Apache is clear the full license text must accompany all derivatives.

So is OFL. Your point? 

A.

[Adam Twardoch]

Of course I'll agree that with both the previous Apache 2-licensed Noto fonts or the current OFL-licensed Noto fonts, and with the majority of the Google-hosted OFL fonts which do not contain the full license text inside the font's "name" table, any 3rd party that distributes such fonts in loose form (potentially including e.g. the dynamic subsets served by Google Fonts or Adobe Typekit etc.) potentially breaks the license. Again, this is independent of whether it's Apache or OFL. 

Actually, OFL is *much more* hardcore: it mandates that “each copy contains the above copyright notice and this license”, so if the font is redistributed loose, then the font must *contain* the license. 

Apache 2 is much more relaxed: it states “You must give any other recipients of the Work or Derivative Works a copy of this License” but it does *not* mandate *when* I must give it to them or in what form. 

With Apache 2, I can give the recipients of the work a copy of the license in any form, and at any time (much before or much after the distribution of the work). 

Apache 2 does not say at all that the license must accompany the work — only that I must give a copy of the license to the recipient. Which means that for if the recipient gets 1000 Apache 2-licensed works from me, just one copy of the license, "a copy" will suffice. 

OFL on the other hand requires that *every copy* of the work *contains* the license. 

Apache 2 makes it about recipients (which effectively limits the maximum number of copies of the Apache 2 license that needs to exist at 7 billion ;) ), OFL makes it about copies of the work, so it's potentially infinite. 

Piece by down, down by piece. :) 

A.

Sent from my mobile phone.

--
You received this message because you are subscribed to a topic in the Google Groups "noto-font" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/noto-font/NYzitp0ONyc/unsubscribe.
To unsubscribe from this group and all its topics, send an email to noto-font+...@googlegroups.com.

To post to this group, send email to noto...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/noto-font/A99AFE08-B31C-4C12-8913-CBA34F7091C7%40twardoch.com.

[Adam Twardoch]

To put it more clearly:

If Google gave me a copy of the Apache 2 license 15 years ago, then it has fulfilled the requirement for any Apache 2-licensed software that Google will ever deliver to me until the day I die.

But with OFL, every copy of every OFL-licensed font that I receive from Google must contain the license — because in Apache 2, the "copy of the license" requirement is tied to the user, while in OFL it's tied to each copy of the work.

It's right there in the licenses, just read them :)

A.

[Adam Twardoch]

The Apache 2 license may include a few nuisances, but it's a very solid piece of legal code, well-written, unambiguous, well thought-out, where every word has been crafted to perfection and tested in countless scenarios worldwide.

OFL may be superficially sinpler and more font-specific. But it's not nearly as good. It's closer to the Python code I write than to the code Behdad would write ;) Of course it's not the terrible mumbo-jumbo the LPPL (LaTeX license on which the GUST license is based). But it's far from perfect.

Dave knows this, but do you jnow that by law, strictly speaking, the OFL license is valid in Poland for only five years after publication and then becomes void, because Polish copyright law requires the term of the license to be stated explicitly, otherwise it defaults to five years. Apache includes the magic word "perpetual", OFL does not.

And this was a real case: Łukasz had to sign an extension for the Polish government before they could use Lato on the Polish ID cards and government forms.

OFL may include more such traps because it hasn't been as thoroughly scrutinized, tested and contested. I could list a few more potential caveats right away: it also doesn't contain the magic word "irrevocable" :)

On the other hand, Apache 2 gives rock-solid legal stability.

So, there is a flurry of arguments that I've presented here. In danger or prospect of becoming that yelling guy who stands on the street corner each day holds some "doom is near" sign, I'll try to shut up now. :)

A.

[Dave Crossland]

Hi Adam

You have written over 6,500 words on this over the weekend, which could probably suffice for a Masters dissertation. Well done! :) I'm not going to go piece by piece, but rather I've attempted to respond to the substance of your arguments.

First, some rather tangental observations:

Your analogies about shoes and yoghurt and Catholics are fun, but I don't think they are helpful because font licensing is what it is, and such comparisons invite cloudy thinking and poor conclusions.

Your definition of license compatibility is new to me, and I think was just made up by you this weekend. I'm familiar with the term 'license compatibility' as meaning two different libre licenses which can allow for a combination of works under each license. I've never seen it used to mean a set of works licensed under the same license! :) And, per your section about FreeType licensing, you also seem to have a unique interpretation of 'sublicensing.' 

Reuse In Restricted Licensing 

Is Noto more valuable if Noto font data can be reused and improved in restricted license fonts? 

After having spent the last 2 years working with dozens of designers on non-latin fonts, I don't think so. The kind of direct use you describe isn't actually very useful in practice. 

The 'script grammer' knowledge is not reduced under OFL. All of that knowledge is just as available under OFL as Apache. Anyone can look at it, study it, and do their own work based on that knowledge. 

Designers prefer to do their own work, so they understand it, instead of blindly copy and pasting code that works magically that they don't understand. This isn't "survival of the fittest," but the actual reality of type designers, not your hypothetical imaginary ones. And that isn't ignoring Noto because it is binaries only; this applies to other projects with full sources available too.

You say that Noto provides practical insights into the world's writing systems: "glyph repertoires relevant for each given script, to some extent the glyph naming, the structure of which glyphs are used as components in composite glyphs, the positioning of the components, the mark attachment anchors, the GSUB OpenType Layout features."

Most, perhaps even ALL of that, can be reused. Afterall, if you cannot take any part of an OFL font (say the GSUB table) and put them into a restricted font, how then do you explain that you and Łukasz Dziedzic could make and sell to Google the Carlito fonts, the version of Lato that is metrics compatible with Calibri, and which therefore copies all the above? The answer may be that some things are not subject to copyright because they are functional, and making functionally-equivalent replacement parts can't be obstructed by bogus copyright claims. 

Noto Symbols could work like the symbols font in the FDK, but how many people use that 'generic' symbols font? Again, in the real world, designers like to draw their own per project, or to use their own 'generics' across their projects. 

Adobe uses the OFL for its libre fonts, and such app developers are just as able to use a Noto GX as a non-latin fallback system under OFL as Apache.

Dual licensing under Apache + OFL is a very poor choice:

Dual licensing is a problem precisely because it is not a problem for the copyright holder, and _only_ not a problem for them. 

For everyone else it adds at least complexity, and at worst legal ambiguity and risk.

For example: You say that when software is dual-licensed, recipients can choose the terms under which they want to use or distribute the software, but the OFL could be interpreted to say that if you receive the fonts under the OFL and another license, then you must choose ONLY the OFL, over that other license, when redistributing. 

I'm not offering advice on if that is what the OFL says, but pointing out that dual licensing adds unnecessary ambiguity and complexity, and _that_ is a much more substantial problem than the problems you purport to address.

You say that "you could always ask for permission" to receive a copy under another license, or for clarification, but the whole point of licensing is to avoid all the asking and answering. Its very expensive! 

Overall, if our goal is to maximize engagement, then I think dual licensing would not be a good strategy to meet that goal. 

Apache is a poor choice:

Doug's first post explained that while the Apache license is "solid" and "crafted to perfection," it turns out to be a bad choice for fonts. 

It is a long text, and while it can be applied to various works, it was written with software programs in mind, unlike OFL. 

You can talk yourself into saying that the problems reported to the Noto team so far were not "real" problems, but that is denying the actual, lived experience of people using Noto. 

Anyone other than the Apache Foundation can write a long FAQ claiming whatever they want, such that distributing parts of the font is 'program output,' or whatever, but how many legal teams who have their own experience and interpretations of the license will agree with what is in such a FAQ?

Changing the working of the notice would be an especially poor choice: Sneaking any extra terms into the License metadata field creates even more complexity and ambiguity. The license should be all in one text, not strewn about all over the place. 

Behdad already said that using Apache and expanding on the ways of telling everyone that the intent is different from the license has already been considered and discarded. I think this is a good choice.

OFL is a good choice:

The "hanging in a museum" problem doesn't exist with OFL because of §5. These two licenses are really quite different on this point! You repeat this point a few times, but really, PDFs with OFL licensed fonts embedded do not need to include a complete copy of the license. 

You keeping saying that "every copy of every OFL-licensed font that I receive from Google must contain the license" but this is not true. You can read SIL's OFL FAQ for confirmation of this, but I will add that a crucial part of §2, that you left out totally, is conveying the license text in a "appropriate machine-readable metadata field" is acceptable; therefore filling the NAME table ID 14 with the URL where the full OFL text can be obtained may be sufficient.

Your anecdote about Polish government lawyers is interesting; I think the government got bad advice, because I think that each time the fonts are distributed to someone, they get a new license, so the 5 year default doesn't matter. I agree that the OFL could be improved, but, I agree with Behdad, inventing a new license doesn't seem like a good use of anyone's time. 

Maximizing Engagement

You claim that professional engagement in copyleft projects is less than non-copyleft projects, but there are many projects with both licenses that fail to engage, and that succeed. In fact the engagement is not only predicated on the license choice, but many other factors. You suggest fontTools sees more engagement because it isn't copyleft, but how much engagement was it seeing in the years before Behdad began actively developing it?  Perhaps no one contributed to FontShop's FontQA because it required FontLab 5, which has been an obvious dead-end for years, and the UFO tools are so active because they were widely seen as the most promising way out of that dead-end? ;)

Wikipedia sees a lot of contribution to its copyleft corpus without contributors being paid. About a dozen libre encyclopedia projects were being tried when Wikipedia started, and they all failed. So did the subsequent projects that attempted to pay people. Fonts - OFL or Apache or GPL or MIT or anything else - very rarely see any contributions. The license just isn't all that relevant for what makes the level of contribution take off.

You say here that you won't contribute to Noto if it is OFL, but I'm not convinced: You recently said on another forum that you have recommended OFL fonts to some of your publishing clients, and adjusted those OFL fonts to your clients needs. You also licensed Lato under OFL, not Apache, when given the choice. 

Finally, Monotype, the principle contributor to the Noto project, prefers that Noto be OFL. 

Conclusion

Just like Monotype, Adobe, Microsoft, Tiro and all the dozens of smaller designer's I've commissioned fonts from - including you and Łukasz! - I think that when releasing libre fonts, OFL is the best choice, and I'm very happy that Noto is now part of the OFL ecosystem. 

Cheers

Dave

[Adam Twardoch (List)]

Dave, 

My apologies to those who are overwhelmed with my >6,500 words and think they’re spam. I might add that I've written half of that thread on my mobile phone. ;) Which, I hope, shows that this is simply an issue I really care about. 

Dave, 

Thank you for your summarized reply. At this point, I don’t think I’m able to add anything substantially new on the subject. 

I will, however, say that you and I differ when it comes to quite a few of the issues you refer to in your reply, but I won’t bother the others here with addressing each of them. I’ll just make a few clarifying points: 

1. I’m not saying the more permissive Apache 2 is *generally* superior to the more restricted OFL, and I wouldn’t support the creation a massive library of Apache 2 fonts, or switching all existing OFL fonts to Apache 2. OFL, being a restricted font license, has its merits. 

OFL obviously allows dual licensing, something that ParaType (PT Sans/Serif), Huerta Tipografica (Alegreya), Monotype (Arimo, Tinos), TypeTogether (Bree) and other professional foundries have made use of, releasing some fonts under two or more restricted licenses: OFL plus their own EULAs. 

Would you care to explain how “OFL + a foundry EULA” (as in the examples given above) is fundamentally different from “OFL + Apache 2”? 

Those “OFL + foundry EULA” dual-licensed fonts are usually two separate (sometimes identical, sometimes slightly different) releases that are typically made under slightly different names (sometimes one having a “Pro” suffix). Yet >95% of the licensed “font software” is still identical between these releases.  

If you think that model is most sensible, then I’ll be most happy if Noto got the two already-existing branches under OFL (“hinted” and ”unhinted”), plus a third ”devel” branch released under Apache 2. I realize that it would be more work for the team, but I think at least some of the arguments I presented in my 6,500 words are convincing.  

2. I’m only advocating for keeping Apache 2 for *Noto* specifically, due to its very unique nature, not really comparable with the bulk of OFL fonts, and because the Noto fonts *are* already licensed under Apache 2. 

I know that Google spent very significant money and personal effort for the Noto project, and if it wasn’t originally released under the permissive terms, I wouldn’t dare to ask Google to liberate Noto. I would be happy that even a restricted (OFL) family of such scope exists. 

I’m only asking to reconsider this decision because I feel like after having tasted caviar, it’s hard to accept eating just potatoes for the rest of your life. 

If future improvements and new scripts of Noto will *only* be released under OFL, this will invariantly lead to two different forks. 

I may try to keep the permissive fork of Noto as ”Toto”, and try to maintain it, porting the non-copyrighted portions from the future restricted Noto over to the permissive Toto (as per Dave’s suggestion). And I may try to find externeous funding to keep the Apache 2 version alive. 

This will be a considerable effort, will lead to unfortunate duplication and dilution, but so be it. It’s important enough to try. 

3. I wouldn’t have gotten involved in creating a metrically compatible font without extensive consultations with different relevant parties, discussing that specific project. 

I did not have certainty that such a project could be done, and only after many conversations I arrived at the conclusion that this project “while not seen with enthusiasm, it would be tolerated”. This is *very* different from having a project where knowledge sharing and reusal is appreciated or perhaps even encouraged. 

It still isn’t generally clear which portions of digital fonts are covered by copyright and which aren’t. Reusing aspects of restrictively licensed fonts requires *much* more care and effort, and individual consultations — so yes, as per your own words, it’s expensive.  

A permissive license gives you more stability that what you’re doing is “fine”, or even “great”. I’d love it if there was at least one ambitious project that would be in that space. 

Which is why, in case of Noto, I’m advocating for Noto to remain under Apache 2 (as one of the licenses at the very least). 

Noto has in my view a rather unique role and purpose. I see it as very different from Stalinist One or all the other wonderful OFL projects. I see Noto as, potentially, “*the one* reference implementation” of Unicode+OpenType script-specific solutions, which is why I’d love for *one*, just *one* project of such ambitious scope to be available under a liberal, permissive license. 

Dave, 

since, at some point in this thread, you’ve decided to enhance this thread by adding a personal flavor to it, publicly citing specifics of past private negotiations between us, referring to specific projects, and — most amusingly — making colorful, though fictitious, references to my sandbox childhood, I hope you’ll accept me also saying this: 

I get the notion of the OFL shared pool, unification etc., all those things that *you* keep advocating. I get that you, Dave, have a particular agenda and a set of values that may, to some extent, be different from mine, or from the likes of mine. 

I get that with the budgets behind Google Fonts, and the Google marketing machine, you can bring many OFL creations to the market, and are able to use your power to push certain people and developments into certain directions. 

This is often a good thing, though that pushing occasionally feels like strong-arming (a certain contract clause seems to be your favorite stick towards others). 

Well, enjoy that power while it lasts, mate. ;) 

It is true that I do recommend OFL fonts for certain projects, and I had, in the past, contributed to or had been involved with, certain OFL projects, despite knowing that it limits and restricts my own options regarding these fonts. 

As you know, I’ve advocated long and extensively to certain people that they should switch from a different restricted opensource license to OFL (though I’ll refrain from mentioning the specifics). 

However, I was hoping that the Google i18n team, that brings the wonderful Noto project to life, would have a slightly different set of goals. 

Goals that would not exclude the needs of the admittedly much smaller community of people like myself — of those “happy few” who hope that in addition to a large *quantity* of OFL fonts (undoubtedly useful for public mass-market consumption, and highly appreciated by many despite the license restrictions), there would be at least *one* high-*quality* family under a permissive license. 

And that such a family, in addition to serving immediate end-user needs, could bring some longer-term benefits that we may now predict quite yet, and therefore we might choose not to restrict the opportunities. 

From my point of view, having 9,000 restricted proprietary-licensed families available for sale elsewhere, 999 restricted OFL-licensed families brought by Google Fonts, and 1 permissive Apache 2-licensed family (Noto) was the perfect world. Just *1*. 

Noto was something that for many years I had hoped would come into existence in my lifetime, and when it had, it was a cause for huge personal celebration for me. 

Noto’s full switch from Apache 2 to “OFL only” is of course *not* the end of the world for me. I am, after all, left with the old state of Noto, which still is fantastic. 

But knowing that the sunny days have ended on September 23, 2015, that we’re back to cloudier days, and that my “Toto” fonts will soon be obsolete or outdated — does not fill me with joy. Because the difference between 9,999 restricted + 1 permissive families, and 10,000 restricted + 0 permissive families only *looks* small. 

Your argumentation has surprised me at times. In one paragraph, you’ve implied that there’s little point in having an Apache 2 FAQ, and then you’ve pointed me to the OFL FAQ in the next one. 

As for getting regular inquiries about the Noto licensing terms — I can assure you that I’m getting very regular questions and doubts about the Lato licensing, which is OFL, and for which a FAQ exists. I may be naive but I don’t think questions and doubts will go away, regardless of the license used. 

I definitely agree with you, Dave, on two points: 

1. You say “Monotype, the principle contributor to the Noto project, prefers that Noto be OFL” — of that I have absolutely no doubt! :) 

2. You win. Good for you! :) 

To the Google i18n team: 

Thank you guys for this wonderful “liberal Noto” trip! It was a great several years-long honeymoon. :) 

It may have ended for me, but I’ll have great memories, and my Toto repo. 

Until better days! :) 

Cheers,

Adam

Sent from my mobile phone.

--
You received this message because you are subscribed to the Google Groups "noto-font" group.
To unsubscribe from this group and stop receiving emails from it, send an email to noto-font+...@googlegroups.com.

To post to this group, send email to noto...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/noto-font/CAEozd0wJcTBsZNp3JWXYCc7MDdVmYLcFYCB1aHF0BZeNHuUqsw%40mail.gmail.com.

[Dave Crossland]

Dear Adam

Lest I give the impression that I have a role in deciding any of this due to my consultancy work with Google, I feel I ought to clarify that I don't have such a role: what the Noto team will decide to do, I'm involved as much as you are - by making reasoned proposals and counterproposals here on this list, as Behdad requested. 

I don't think licensing source files under Apache and binaries under OFL is a good idea. If one was to take Apache - or BSD/MIT - licensed font works and combine them with original works and distribute the whole under the OFL, I am not sure that this would be possible - because of the OFL's "no other license" requirement. Perhaps the only non-OFL license compatible with OFL is that other rather odd one that you obliquely reference, which specifically allows for redistribution under other licenses. 

I'm not offering an opinion on if OFL can or can not be dual licensed in the way you describe. I am pointing out that there is ambiguity around it, especially downstream for end users, so that even if it is possible, it is not a good choice.

“OFL + a foundry EULA” is different from “OFL + Apache 2” because the former has the two sets of fonts (even if nearly bitwise identical) distributed under 2 licenses separately, not a single set of fonts distributed under both licenses where the recipient has to choose between the two. 

Regarding FAQs, I think its wise to distinguish between 1st and 3rd party FAQs: SIL maintains the SIL OFL FAQ to increase the clarify of their intent as the license author. That's different to having anyone other than the Apache Foundation write an Apache licensing FAQ, which I think reduces clarity because it is "telling everyone that our intent is different from our license."

--

Cheers
Dave

[Adam Twardoch]

Dave, 

you were a bit ambiguous in terms of “who’s the 1st and who’s the 3rd party”, so I’ll respond with my point of view. 

Even if software is published under a “standardized” license like OFL or Apache 2, it is still the *copyright owner* of the published software who is the *1st party*, and the recipient of the work is the 2nd party. 

The authors of the standardized license *are* a 3rd party. Still important, but much less important than the copyright owner, because *only* the copyright owner is entitled to actually take some action, should he feel his right infringed by some user. 

Some users will refer to FAQs provided by the authors of the standardized license (who more likely have the legal expertise, and who often craft the FAQs based on *many* inquiries), but in case of particularly popular projects, many users may want to seek clarification from, or refer to FAQs provided by the copyright owner of the particular work that is being licensed. I wouldn’t discount either FAQ — they can be both equally informative and are both equally non-binding. 

After all, if a user breaks the Apache 2 or OFL license with Noto, it is only Google who might sue that user, never the Apache Foundation or SIL. 

Of course if *I* wrote a FAQ for interpreting Apache 2 in context of the Noto fonts, I would be a 4th party at best, so that FAQ would just be a set of purely hypothetical, speculative suggestions. 

Adam

Sent from my mobile phone.

--
You received this message because you are subscribed to a topic in the Google Groups "noto-font" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/noto-font/NYzitp0ONyc/unsubscribe.
To unsubscribe from this group and all its topics, send an email to noto-font+...@googlegroups.com.

To post to this group, send email to noto...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/noto-font/CAEozd0xiPWXiCrmh0gKYybA5D5TU7SErmONofGo4OncYN0mtSw%40mail.gmail.com.

[Dave Crossland]

Dear Adam

Ah, yes, right, I can see how 1st/3rd party can be defined differently :)  

I understand that you see the 1st party as the licensor/work-author, and the license-author as a 3rd party, but I think this isn't wise.

The issue at hand is the intent of the license-author. 

As work-authors, when we used 'standardised' licenses, we abdicate the intent of the license to the license-author; our intentions shift from "what we say" to "what they said."

A FAQ from the license-author clarifies their intent. 

So on this issue, I see the license-author as the 1st party, the licensee as the 2nd party, and the licensor/work-author and anyone else as 3rd parties. 

Also, I believe that the FAQs of license-authors do actually have some binding powers, but FAQs from work-authors don't, and just muddy the waters, and are not a good idea.

--

Cheers
Dave