[x3d-public] draft X3D 4.1 prose for font files and libraries
John Carlson
yottzumm at gmail.com
Tue Mar 4 14:20:11 PST 2025
My vote is with FontStyle.url too! One has to deal with elegant design
versus straightforward design. My vote is most always with straightforward
design, despite future issues. We can always add FontLibrary later, if
needed. Is there any leftover issues not handled by FontStyle.url? Just
font collections? Can we provide or suggest utilities to break apart font
collections?
John
On Tue, Mar 4, 2025 at 4:11 PM Michalis Kamburelis via x3d-public <
x3d-public at web3d.org> wrote:
> To make my questions about FontLibrary easier to understand, I prepared
> the testcases, they are in
> https://github.com/michaliskambi/x3d-tests/tree/master/font_library_questions
> . They contain 4 questions in total -- 2 about Inline, 1 about searching
> non-top-level , 4th about EXTERNPROTO.
>
> You can git clone the whole repo of
> https://github.com/michaliskambi/x3d-tests/ and go into subdirectory
> font_library_questions to get all these files easily.
>
> Please don't take the fact that I made theses tests as my encouragement to
> pursue the FontLibrary design :)
>
> - I'm really more and more convinced that FontLibrary is an unnecessary
> complication. Given that font collections are rare (and can be supported in
> simpler way), and opentype.js doesn't support them... I really think this
> is an over-complicated design -- unnecessary for users, and possibly not
> ever fully implementable by any browser.
>
> - And in contrast, "FontStyle.url" seems like something totally obvious. 2
> browsers already implement this, or something very close, from what I
> understand. (And for my browser it is also trivial, so we have 3
> implementations that confirm it's OK.) And all my 4 questions (from
> https://github.com/michaliskambi/x3d-tests/tree/master/font_library_questions
> ) just disappear in the "FontStyle.url" version. Associating the font file
> to FontStyle is explicit and obvious in this case.
>
> Anyhow, I repeated my sentiment enough times already :), I'll let others
> speak. From the spec standpoint, we need at least answers to my 4
> questions, to understand the edge-case behavior.
>
> Regards,
> Michalis
>
>
> wt., 4 mar 2025 o 19:08 Michalis Kamburelis <michalis.kambi at gmail.com>
> napisał(a):
>
>> Don,
>>
>> No worries, I will give you time then to carefully read my mail :)
>>
>> In AD 2 I raise some questions still unanswered ("""If
>> FontStyle.fontLibrary = NULL, then match the family name from
>> FontStyle.family, using all loaded fonts from FontLibrary nodes. (Should it
>> take into account also FontLibrary nodes mentioned only in some
>> FontStyle.fontLibrary? or only top-level FontLibrary nodes? and what about
>> my questions with Inline and EXTERNPROTO cases?)"""). More explanation
>> about Inline and EXTERNPROTO questions in previous mail in this thread.
>>
>> We also miss more testcases. Testcases that address my above questions,
>> showing what happens in the cases I question about, would be desirable. We
>> also miss example TTC, OTC, multiple-family WOFF and then testcases of X3D
>> using them.
>>
>> As for your argument """overloading FontStyle with X3DUrlObject
>> functionality is a bad idea, and do not support it. Presentation ought to
>> be separate from network-loading functionality.""", I admittedly don't
>> agree with it. Possibly, I don't understand where this argument is coming
>> from :) We add X3DUrlObject to some presentation things, it's not a problem
>> in my view. E.g. ImageTexture is X3DUrlObject, just like all other texture
>> nodes, and AudioClip, and Inline... Adding X3DUrlObject to FontStyle would
>> be consistent, in my view.
>>
>> As for implementations:
>>
>> - For X3DOM and X_ITE, from what I understand, implementing
>> "FontStyle.url" is not a problem. I understand that X_ITE already had a
>> similar functionality (just over-using FontStyle.family for URLs), and
>> X3DOM already implements experimental FontStyle.url. Adding FontStyle.url
>> to Castle Model Viewer/Castle Game Engine would be trivial too.
>>
>> - This is in contrast to FontLibrary. Andreas says """x3dom (or any
>> browser) will not be able to support font collections""". Desktop Castle
>> Model Viewer/Castle Game Engine... I don't know can I support it either
>> (FreeType can query font collection by a known index, I don't see how to
>> enumerate all fonts inside, and implement the spec would require it).
>>
>> Regards,
>> Michalis
>>
>>
>>
>> wt., 4 mar 2025 o 16:35 Brutzman, Donald (Don) (CIV) <brutzman at nps.edu>
>> napisał(a):
>>
>>> Thanks for careful review Michalis.
>>>
>>> Dick Puk and I met Monday (before your email) and made some refinements.
>>>
>>> - Scrapped 'enabled' field
>>> - Improved prose, closely reviewed Examples 1 and 2.
>>> - We have been carefully tracking the email threads to date, and
>>> think we have addressed almost all concerns.
>>>
>>> Some of our prose refinements match yours, some don't.
>>>
>>> - We really think that overloading FontStyle with X3DUrlObject
>>> functionality is a bad idea, and do not support it. Presentation ought to
>>> be separate from network-loading functionality. That overstrike-cyan
>>> prose has been removed.
>>> - As a result, each of your FontStyle.url examples remain supported
>>> by FontLibrary.url field.
>>> - Now that Example 1 (FontLibrary outside of FontStyle) and Example
>>> 2 (FontLibrary inside of FontStyle), few advantages appear relevant for
>>> adding a FontStyle.fontLibrary field.
>>> - This addresses many of the complications that you point out.
>>>
>>> - X3D 4.1 (Draft) Architecture, clause 15.4 Node reference
>>> -
>>> https://www.web3d.org/specifications/X3Dv4Draft/ISO-IEC19775-1v4.1-CD//Part01/components/text.html#NodeReference
>>>
>>> Summary of our review:
>>>
>>> - it seems like the current two-path approach offers compatible
>>> alternatives, and avoiding FontStyle.fontLibrary field completely still
>>> works and is simpler... Thus preferred.
>>> - When multiple different font families exist within a single font
>>> file, browser inspection of name information inside that file seems
>>> necessary.
>>>
>>> Michalis, I will study your email further when time permits. This is a
>>> preliminary response describing yesterday evening's specification efforts.
>>>
>>> Just ran some more tests. Progress report: it is super impressive that
>>> both EXAMPLE 1 and EXAMPLE 2 run satisfactorily in both X3DOM Editor and
>>> X_ITE Playground... A real testament to great implementations that
>>> variations to expectations are supported. Standing Ovation!
>>>
>>> Further improvements are welcome, of course. However we have come a
>>> long way... Am hoping to avoid a spiraling email cyclone.
>>>
>>> If we are converging, great. Incidentally it does not hurt to keep both
>>> of these approaches and work further on examples implementations.
>>>
>>> If we are clarifying alternatives, yes please let's keep improving draft
>>> specification prose.
>>>
>>> If we are not converging, that is OK too... we might consider having a
>>> meeting to discuss with each other, to understand goals/contraints/problems
>>> and example use cases well. "First seek to understand" seems apropos.
>>>
>>> Onward we go... Again thanks for much excellent thinking and multiple
>>> great efforts.
>>>
>>>
>>> all the best, Don
>>>
>>> --
>>>
>>> Don Brutzman Naval Postgraduate School, Code USW/Br
>>> brutzman at nps.edu
>>>
>>> Watkins 270, MOVES Institute, Monterey CA 93943-5000 USA
>>> +1.831.656.2149
>>>
>>> X3D graphics, virtual worlds, navy robotics
>>> https://faculty.nps.edu/brutzman
>>>
>>>
>>>
>>> ------------------------------
>>> *From:* Michalis Kamburelis <michalis.kambi at gmail.com>
>>> *Sent:* Tuesday, March 4, 2025 4:43 AM
>>> *To:* Andreas Plesch <andreasplesch at gmail.com>
>>> *Cc:* Brutzman, Donald (Don) (CIV) <brutzman at nps.edu>; Holger Seelig <
>>> holger.seelig at yahoo.de>; Extensible 3D (X3D) Graphics public discussion
>>> <x3d-public at web3d.org>
>>> *Subject:* Re: [x3d-public] draft X3D 4.1 prose for font files and
>>> libraries
>>>
>>>
>>> NPS WARNING: *external sender* verify before acting.
>>>
>>> 1. I looked how FreeType supports font collections.
>>> https://freetype.org/ indeed says it can handle TTC, OTC.
>>>
>>> API: we can pass "index" to various functions (like FT_New_Face),
>>> which seens to be an index into the font collection.
>>>
>>> I didn't find any way to query the available count inside a given
>>> font collection (file) -- it seems that application is just expected to
>>> know the index, i.e. it is specified by the user like "use font from this
>>> amazing-font.ttc file, index 1". It seems to me that "enumerate all fonts
>>> within a font collection" is not a typical operation, I actually don't know
>>> how to do it yet.
>>>
>>> 2. I am afraid we go into something overly complicated with FontLibrary,
>>> both to use (for authors) and to implement.
>>>
>>> The current design of FontLibrary assumes that font collections
>>> (that include many families and/or many font styles within a single file)
>>> are a common use-case. But this is not the case, it seems we all confirmed
>>> this in this thread. In practice, font files are TTF / OTF / WOFF with one
>>> family and one style. And opentype.js does not support font collections.
>>> FreeType suppprts them, but I'm not sure how to enumerate them all.
>>>
>>> The new prose accounts for this, by allowing various options... and
>>> being needlessly complicated, IMHO. So I feel we arrived at a complicated
>>> spec, without much benefit.
>>>
>>> If I understand the new proposal (
>>> https://www.web3d.org/specifications/X3Dv4Draft/ISO-IEC19775-1v4.1-CD//Part01/components/text.html#FontLibrary
>>> ) right:
>>>
>>> The browser should read the first accessible font file from
>>> FontLibrary.url.
>>>
>>> This font file, referenced by FontLibrary.url, can be:
>>>
>>> - Multiple font families inside. If the first accessible thing on
>>> FontLibrary.url is a font collection (TTC, OTC, or a collection inside
>>> WOFF). And then "FontLibrary.family" cannot be used (as it would make no
>>> sense).
>>>
>>> - Single font family (with multiple styles or one style). And then
>>> (according to spec) "FontLibrary.family" is allowed (to override family
>>> name from font file content), but optional.
>>>
>>> The connection between FontLibrary and FontStyle can be done in 2
>>> ways, as I understand:
>>>
>>> - If FontStyle.fontLibrary = NULL, then match the family name from
>>> FontStyle.family, using all loaded fonts from FontLibrary nodes. (Should it
>>> take into account also FontLibrary nodes mentioned only in some
>>> FontStyle.fontLibrary? or only top-level FontLibrary nodes? and what about
>>> my questions with Inline and EXTERNPROTO cases?)
>>>
>>> - If FontStyle.fontLibrary <> NULL, then match the family name from
>>> FontStyle.family, using only the indicated FontLibrary node.
>>>
>>> This is also quite different to what CSS is doing. At least to what
>>> I shown in
>>> https://stackoverflow.com/questions/12812441/how-do-i-use-woff-fonts-for-my-website
>>> .
>>>
>>> 3. I really propose we go back to "FontStyle.url", and scratch the
>>> "FontLibrary" node design.
>>>
>>> We can still add optional support font collections (noting that
>>> opentype.js cannot implement this), but allowing a trivial `#<index>` at
>>> the font file URL. Like
>>>
>>> FontStyle {
>>> url "my-font-collection.ttc#1"
>>> }
>>>
>>> And in the most common case, it is simple:
>>>
>>> FontStyle {
>>> url "my-font-file.ttf"
>>> }
>>>
>>> Advantage of FontStyle.url: The common use-case, which is: font file
>>> is one family with one style, remains now trivial. All my questions from AD
>>> 2 disappear :) And all implementations can easily support it.
>>>
>>> Regards,
>>> Michalis
>>>
>>> pon., 3 mar 2025 o 21:00 Andreas Plesch <andreasplesch at gmail.com>
>>> napisał(a):
>>>
>>> Thanks, I think that captures the design ideas well.
>>>
>>> I think font collection files are very rare. They are explicitly not
>>> supported by the opentype.js library (
>>> https://github.com/opentypejs/opentype.js/discussions/668). But they do
>>> exist and are supported by the freetype library. So it could become
>>> optional for x3d browsers to support those or not, perhaps via component
>>> levels. There seem to be many tools which can split a font collection file
>>> into separate, single font files.
>>>
>>> Following CSS, one idea to deal with font collections (.ttc, .otc files,
>>> also as embedded in woff) would be to require an author to select a
>>> specific, single font from a collection via url# fragment syntax:
>>>
>>> url=' "
>>> https://server/path/largeExoticFontLibrary.ttc#Postscript_name_of_FontA"
>>> '
>>>
>>> CSS uses the Postscript name (format defined in a RFC,
>>> https://www.rfc-editor.org/rfc/rfc8081#section-4.2) for selection
>>> according to the CSS spec.
>>>
>>> That way font collection files can be treated the same as single font
>>> files. I do not know if or how freetype manages font collection files.
>>> Also, I do not know if the CSS syntax above is supported by web browsers
>>> (but I would expect them to,
>>> https://bugzilla.mozilla.org/show_bug.cgi?id=1529652 says no).
>>>
>>> For global FontLibrary scope, in my opinion, the FontLibrary.family
>>> field should not be optional but required, again following the CSS pattern.
>>> The family name is easy for an author to add, and makes it unambiguous for
>>> a x3d browser to match, without first downloading the file for inspection.
>>> This then would mean that an empty family field does not guarantee a match
>>> although a browser may use the internal font name(s) as a fallback.
>>>
>>> With the url selection mechanism above, a global FontLibray node may not
>>> be necessary at all. But it remains questionable if there is enough library
>>> support for font collection files in general.
>>>
>>> -Andreas
>>>
>>>
>>> On Mon, Mar 3, 2025 at 12:26 PM Brutzman, Donald (Don) (CIV) <
>>> brutzman at nps.edu> wrote:
>>>
>>> Excellent analysis together is much appreciated.
>>>
>>> To facilitate continue discussion on this highly detailed discussion, I
>>> have a applied some additions to the "design draft" specification prose:
>>>
>>> - Add FontLibrary family field
>>> - SFString [in,out] family "" (under consideration)
>>> - The *family* field provides an identifying name for the
>>> associated font library. This field provides a reference label for the
>>> *family* field in a corresponding FontStyle or ScreenFontStyle
>>> node.
>>> - (Editors note: this approach is similar to how CSS can identify
>>> font family for use in HTML.)y
>>> - (Editors note: key design question for whether global scope is
>>> needed: can a single font file contain more than one font?)
>>> - Added description for EXAMPLE 1.
>>> - FontLibrary node with global scope, where font family
>>> identification is deduced from name metadata within the font file.
>>> - Added second example, adapting Holgers design-exploration
>>> example
>>> - EXAMPLE 2. FontLibrary nodes each having local scope within parent
>>> FontStyle node, where font family identification is provided by matching
>>> values for corresponding *family *fields.
>>> - # example derived from FontLibrary.x3dv
>>> <https://create3000.github.io/preview/playground/?url=https://gist.githubusercontent.com/create3000/6d3992dea7e641bb59645684ca468ce7/raw/a63e66e23b305478fbfdee9e67d7db9cad1f8e72/FontLibrary.x3dv> (Holgers
>>> example)
>>> - Moved FontLibrary nodes as *fontLibrary *field in respective
>>> FontStyle nodes, added FontLibrary DEF and matching *family *
>>> fields.
>>> - Screenshot attached - wow, X_ITE rendered it correctly without
>>> further software adjustment! 🙂
>>> -
>>> - X3D 4.1 Draft Architecture, clause 15 Text component, 15.4.1
>>> *FontLibrary*
>>> -
>>> https://www.web3d.org/specifications/X3Dv4Draft/ISO-IEC19775-1v4.1-CD//Part01/components/text.html#FontLibrary
>>> - X3D 4.1 Draft Architecture, clause 15 Text component, 15.4.2
>>> *FontStyle*
>>> -
>>> https://www.web3d.org/specifications/X3Dv4Draft/ISO-IEC19775-1v4.1-CD//Part01/components/text.html#FontStyle
>>>
>>>
>>> Hopefully this is capturing and synthesizing everyone's intent. Further
>>> clarifications and refinement are of course welcome.
>>>
>>> Onward we go, getting clearer through scrutiny during each pass. Have
>>> fun with X3D FontLibrary!
>>>
>>>
>>> all the best, Don
>>>
>>> --
>>>
>>> Don Brutzman Naval Postgraduate School, Code USW/Br
>>> brutzman at nps.edu
>>>
>>> Watkins 270, MOVES Institute, Monterey CA 93943-5000 USA
>>> +1.831.656.2149
>>>
>>> X3D graphics, virtual worlds, navy robotics
>>> https://faculty.nps.edu/brutzman
>>>
>>>
>>>
>>> ------------------------------
>>> *From:* Andreas Plesch <andreasplesch at gmail.com>
>>> *Sent:* Monday, March 3, 2025 4:06 AM
>>> *To:* Brutzman, Donald (Don) (CIV) <brutzman at nps.edu>
>>> *Cc:* Holger Seelig <holger.seelig at yahoo.de>; Extensible 3D (X3D)
>>> Graphics public discussion <x3d-public at web3d.org>
>>> *Subject:* Re: [x3d-public] draft X3D 4.1 prose for font files and
>>> libraries
>>>
>>> Similar to Michalis viewpoint, I also see advantages of a FontStyle.url
>>> (or FontStyle.fontLibrary) field over a global FontLibrary node. x3dom
>>> currently does not need to download, parse or render font files at all
>>> since this can all be delegated to the web browser in which x3dom is
>>> necessarily embedded. It does not currently need a library such as
>>> opentype.js and can take advantage of the rapid pace of web browser
>>> development. For example, x3dom could use variable fonts or color fonts.
>>>
>>> On Sun, Mar 2, 2025 at 6:26 PM Brutzman, Donald (Don) (CIV) <
>>> brutzman at nps.edu> wrote:
>>>
>>> An optional *family* field might be helpful for FontLibrary but
>>> probably only when a single font family is contained in that file.
>>>
>>>
>>> True, a FontLibrary.family field would only be helpful in that case.
>>> However, the vast majority of font files only contain one font family. In
>>> fact, I have not come across any font file which contains multiple families
>>> (tfc,otc). Variable fonts have one family but multiple styles and weights.
>>> My suggestion would be to strongly recommend to authors to provide a
>>> FontLibrary.family name. This guarantees correct matching with a FontStyle.
>>> In fact, x3dom would effectively require use of the field.
>>>
>>> Font files can also be collections according to WOFF2 Recommendation.
>>>
>>> - WOFF File Format 2.0, W3C Recommendation, 08 August 2024
>>> - Status of This Document
>>> - "The WOFF 2.0 specification is implemented in all major browsers,
>>> and is widely used on production websites. It supports the entirety of the
>>> TrueType and OpenType specifications, including Variable fonts, Chromatic
>>> fonts, and font Collections."
>>> - https://www.w3.org/TR/WOFF2
>>> - 4.2. Collection directory format
>>> - "If input font is a collection, the table directory
>>> <https://www.w3.org/TR/WOFF2/#table_dir_format> contains a single
>>> entry for each unique table in the entire collection, regardless of whether
>>> a table is shared among different fonts in a collection or not. Thus, the
>>> table directory for a font collection compressed with WOFF2 comprises all
>>> tables from all fonts contained in the font collection file."
>>> - https://www.w3.org/TR/WOFF2/#collection_dir_format
>>>
>>>
>>> My understanding is that this collection capability is only used by some
>>> Asian fonts where the same glyph can be shared by different font families,
>>> for different characters.
>>> I am not sure how CSS or freetype is currently dealing with this
>>> situation. It should be insightful to find out. Found out something for
>>> CSS, see below.
>>>
>>> We might be able to define this so that it works both ways, either with
>>> matching * family *names or not... for example,
>>>
>>> - a FontLibrary with a woff url referring to a single font might be
>>> an unambiguous child of FontStyle as FontStyle.fontLibrary. This approach
>>> completely avoids any need to check for precise *family *naming
>>> matches., the FontStyle only uses that contained font
>>>
>>> Yes, I think this should work fine.
>>>
>>>
>>> - a FontLibrary with a woff url referring to a collection of fonts,
>>> which are each named and matchable, might also be used for matching with
>>> FontStyle.family.
>>>
>>>
>>> I fear we do not know enough about font files with a collection of fonts
>>> are intended to be used.
>>>
>>>
>>> - DEF/USE declarations allow us to have both global and restricted
>>> availability as desired by the author. In other words,
>>> - a FontLibrary might not be referenced as a field and have global
>>> scope (the prose that is currently in the 4.1 design draft), or else
>>> - a FontLibrary might also be the child of a FontStyle node
>>> (something we might add to the 4.1 design draft).
>>>
>>>
>>>
>>> Seems like we might be able to "take the fork in the road" here, giving
>>> authors flexibility. Hmmm...
>>>
>>> Somewhat related: since staying aligned with CSS practice is important,
>>> here is an example of further interest. The following provides a clear
>>> example of how an author can define CSS so that html can use four different
>>> woff files for plain, bold, bold-italic, italic markup (using <b> and <i>
>>> elements). Food for thought perhaps.
>>>
>>> - How do I use .woff fonts for my website?
>>> -
>>> https://stackoverflow.com/questions/12812441/how-do-i-use-woff-fonts-for-my-website
>>>
>>>
>>>
>>> This is an old but still working and good example.
>>> https://developer.mozilla.org/en-US/docs/Web/CSS/@font-face and
>>> https://drafts.csswg.org/css-fonts/#font-face-rule have more details.
>>> Note how the font-family is a given custom name which is then used in
>>> other css rules.
>>> https://drafts.csswg.org/css-fonts/#font-face-src-loading has the
>>> syntax of how to select a font from a collection, using its Postscript name
>>> (so not the family name or full name).
>>>
>>> All the best, -Andreas
>>>
>>>
>>> all the best, Don
>>>
>>> --
>>>
>>> Don Brutzman Naval Postgraduate School, Code USW/Br
>>> brutzman at nps.edu
>>>
>>> Watkins 270, MOVES Institute, Monterey CA 93943-5000 USA
>>> +1.831.656.2149
>>>
>>> X3D graphics, virtual worlds, navy robotics
>>> https://faculty.nps.edu/brutzman
>>>
>>>
>>>
>>> ------------------------------
>>> *From:* x3d-public <x3d-public-bounces at web3d.org> on behalf of Andreas
>>> Plesch via x3d-public <x3d-public at web3d.org>
>>> *Sent:* Sunday, March 2, 2025 1:39 PM
>>> *To:* Holger Seelig <holger.seelig at yahoo.de>
>>> *Cc:* Andreas Plesch <andreasplesch at gmail.com>; X3D <
>>> x3d-public at web3d.org>
>>> *Subject:* Re: [x3d-public] draft X3D 4.1 prose for font files and
>>> libraries
>>>
>>> Thanks. That is what I thought. But should this mapping be part of the
>>> X3D spec. ? Or should this mapping be left at the digression of the X3D
>>> browser as a best effort ? Eg. most browsers may use fontFamily to
>>> associate a FontStyle.family with but some browsers may prefer to use
>>> fullName and expect FontStyle.family to match that. Should the matching be
>>> case sensitive ?
>>>
>>> I think it would be cleaner to not have to rely on matching font names.
>>> A FontLibrary.family would allow that and not require any changes to
>>> FontStyle as it is defined.
>>>
>>> -Andreas
>>>
>>> On Sun, Mar 2, 2025 at 3:44 PM Holger Seelig <holger.seelig at yahoo.de>
>>> wrote:
>>>
>>> fontFamily - is equal to the FontStyle.family field
>>> fontSubfamily - is equal to FontFamily.style field but can contain
>>> spaces and any kind of chars or even special chars
>>> fullName - could be usefull to directly access a specific font
>>>
>>> This means more or less that the FontFamily.style field should be able
>>> to have more possible values than the current 4 (PLAIN, BOLD, ITALIC,
>>> BOLDITALIC) so that values like "Regular", "Oblique", "Outline", "Outline
>>> Bold" or what ever the Font Designer wishes it should be called and written.
>>>
>>> Best regards,
>>> Holger
>>>
>>> --
>>> Holger Seelig
>>> Leipzig, Germany
>>>
>>> holger.seelig at yahoo.de
>>> https://create3000.github.io/x_ite/
>>>
>>> Am 02.03.2025 um 13:21 schrieb Holger Seelig via x3d-public <
>>> x3d-public at web3d.org>:
>>>
>>> It seems I was wrong with my last email.
>>>
>>> I created a reference implementation of FontLibrary node to test and
>>> play with.
>>>
>>> It is available at:
>>>
>>> https://create3000.github.io/test/?url=https://gist.githubusercontent.com/create3000/6d3992dea7e641bb59645684ca468ce7/raw/a63e66e23b305478fbfdee9e67d7db9cad1f8e72/FontLibrary.x3dv
>>>
>>> Best regards,
>>> Holger
>>>
>>> --
>>> Holger Seelig
>>> Leipzig, Germany
>>>
>>> holger.seelig at yahoo.de
>>> https://create3000.github.io/x_ite/
>>>
>>> Am 02.03.2025 um 05:26 schrieb Andreas Plesch via x3d-public <
>>> x3d-public at web3d.org>:
>>>
>>> Ok, I think these explanations make the intention very clear.
>>>
>>> On Sat, Mar 1, 2025 at 3:29 PM Brutzman, Donald (Don) (CIV) <
>>> brutzman at nps.edu> wrote:
>>>
>>> Thanks for careful scrutiny and questions. Overall responses:
>>>
>>> 1. We have not seen a need for FontLibrary to be referenced in the
>>> field of any other node. It adds to the fonts available in the browser.
>>>
>>> Then a FontLibrary node needs to be able to be a root node:
>>> https://www.web3d.org/specifications/X3Dv4/ISO-IEC19775-1v4-IS/Part01/concepts.html#Rootnodes
>>> So it could be a ChildNode or another class of root nodes.
>>>
>>>
>>> 1.
>>> 2. If someone has a use case where FontLibrary needs to be
>>> specifically referred to by individual FontStyle or ScreenFontStyle nodes,
>>> please state it.
>>>
>>> This would clearly associate a font with an individual FontStyle and
>>> avoid mistakes.
>>> In rare cases there may be two different fonts which have the same
>>> family name. There is no central registry for font family names.
>>>
>>>
>>> 1.
>>> 2. Since fonts themselves change so infrequently, keeping
>>> FontLibrary scope global and not requiring reloads of any given url (local
>>> file or online) seems sensible.
>>>
>>> Rereferencing by DEF/USE a FontStyle.fontLibrary would not require
>>> reloads. In terms of loading efficiency it does not matter if the scope is
>>> Scene or FontStyle/Text.
>>>
>>>
>>> 1.
>>> 2. Since X3DUrlNode functionality of the url (and related) fields
>>> offers so much functionality, making X3DUrlNode an abstract interface for
>>> FontLibrary adds much common functionality found in other X3DUrlNode nodes
>>> that have url functionality (Anchor, ImageTexture, Inline, and others).
>>> This simplifies implementations, keeps functionality consistent, and
>>> provides well-understood semantics to authors. Furthermore, allowing an
>>> author to selectively load a new font file at run time offers greater
>>> flexibility for textual presentation and animation
>>>
>>>
>>>
>>>
>>> 1.
>>> 2. Having url and run-time capabilities means that FontLibrary would
>>> be an X3D node, not an X3D statement.
>>>
>>> True. The statement role just seems to fit better as FontLibray would
>>> not be referenced by any field. Perhaps there is a way to adopt the url
>>> field for such a statement.
>>>
>>>
>>> 1.
>>> 2. The availability of DEF and USE capabilities is perhaps not
>>> needed for regular use. Nevertheless it might provide interesting options
>>> if an author wants to use FontLibrary in combination with Script or
>>> ProtoDeclare/ExternProtoDeclare functionality. For example, a scene might
>>> have a Script or ProtoInstance that uses a specific custom
>>> ExternProtoDeclare reference. Several possibilities for advanced authoring
>>> here, just taking advantage of existing X3D flexibility and extensibility.
>>>
>>> See *response* paragraphs inserted below for specific replies.
>>>
>>>
>>> *response: *We are thinking that if inconsistencies exist between a
>>> specified FontStyle family field and the font file, then such cases are
>>> best handled on a best-effort basis by browsers without further complex
>>> (and possibly contradictory) guidance in specification prose. Perhaps this
>>> stance should be a NOTE?
>>>
>>>
>>> Another source of conflict is the font family name. While most (perhaps
>>> not all) font files contain a correct family name, it is difficult to
>>> extract and requires a library. For example, web browsers do not extract
>>> the family name from a font file, and therefore do not rely on such an
>>> extracted name. An author may not even know the exact font name as
>>> contained in a font file. The family names shown in online font collections
>>> may not match exactly what is contained in a file.
>>> Don's example scene is actually a victim of this potential for
>>> confusion. The family name Playwright_CA_Guides is in reality not the
>>> same name as given in the woff2 file. I think it is Playwright CA
>>> Guides. I may be responsible for this since I used it in an example but
>>> it shows how quickly this can go wrong.
>>> The solution in CSS is to allow authors to define their own family name
>>> for the font given in the font file.
>>> The X3D equivalent would be to add a SFString FontLibrary.family field
>>> which would be the authoritative family name. The family name in the font
>>> file would be ignored.
>>> Then the author can use the same name in the FontStyle.family field.
>>> This starts to sound a lot like DEF/USE names which would work similarly
>>> for a Font Style.font Library field node.
>>>
>>> So please consider adding a SFString FontLibrary.family field to
>>> FontLibrary.
>>>
>>> I think x3dom may not be able to fully implement FontLibrary if
>>> extraction of the family name from the font file is necessary since the
>>> size of the needed libraries may not be commensurate with the added benefit.
>>>
>>> How does FontLibrary define a "collection of one or more font families"
>>> ? Is there an example of a font file containing more than one font family ?
>>>
>>> "An empty *url* list indicates that the default set of fonts provided
>>> by the browser are used."
>>>
>>> The url field always exists but can be empty. So is optional and could
>>> be removed.
>>> Is that case different from not providing a FontLibrary node at all ? If
>>> not, perhaps make this stronger: "list is equivalent to not including the
>>> FontLibrary node."
>>>
>>> Andreas
>>>
>>> P.S. Interesting reference
>>>
>>> - Google Fonts: Using type in AR & VR
>>> - https://fonts.google.com/knowledge/using_type_in_ar_and_vr
>>>
>>>
>>> all the best, Don
>>> --
>>> Don Brutzman Naval Postgraduate School, Code USW/Br
>>> brutzman at nps.edu
>>> Watkins 270, MOVES Institute, Monterey CA 93943-5000 USA
>>> +1.831.656.2149
>>> X3D graphics, virtual worlds, navy robotics
>>> https://faculty.nps.edu/brutzman
>>>
>>>
>>>
>>> ------------------------------
>>> *From:* x3d-public <x3d-public-bounces at web3d.org> on behalf of Andreas
>>> Plesch via x3d-public <x3d-public at web3d.org>
>>> *Sent:* Saturday, March 1, 2025 6:45 AM
>>> *To:* X3D Graphics public mailing list <x3d-public at web3d.org>
>>> *Cc:* Andreas Plesch <andreasplesch at gmail.com>
>>> *Subject:* Re: [x3d-public] draft X3D 4.1 prose for font files and
>>> libraries
>>>
>>> I basically had the same questions as Michalis.
>>>
>>> Comments below.
>>>
>>>
>>> Date: Sat, 1 Mar 2025 05:52:24 +0100
>>> From: Michalis Kamburelis <michalis.kambi at gmail.com>
>>>
>>> As for FontLibrary,
>>>
>>> https://www.web3d.org/specifications/X3Dv4Draft/ISO-IEC19775-1v4.1-CD/Part01/components/text.html#FontLibrary
>>> : I admit I don't understand this proposal (yet).
>>>
>>> 1. Where is the FontLibrary actually used? It doesn't seem you can refer
>>> to
>>> a FontLibrary node in any way. It's also not X3DChildNode, so you cannot
>>> just place it in groups.
>>>
>>>
>>> This is a big missing piece.
>>> After some thought, I think the idea is to give the X3D browser
>>> information on how to render fonts which are referenced in a
>>> FontStyle.family field.
>>> One option is to think of it not as a Node but as a Statement describing
>>> a resource available to the X3D browser in general.
>>> Another option is to make it part of WorldInfo as a value for a new
>>> WorldInfo.fonts field ?
>>> The idea came up as an alternative to a FontStyle.url field, so see
>>> below.
>>>
>>> *response:* See above.
>>>
>>> 2. The sentence """If present, only the first active font library
>>> retrieved
>>> from the url list shall be used."""" --- What is the meaning of word
>>> "used"
>>> in this sentence? Used for what?
>>>
>>>
>>> What is 'active' ? Probably accessible
>>>
>>> "Used": ?
>>>
>>> *response: *We were trying to emphasize that only one font file can be
>>> loaded by a FontLibrary node. The functionality of an X3D url is
>>> unchanged, the first working address in the list is used. This is trying
>>> to complement the more general wording in 9.3.2 concisely... Specific
>>> wording suggestions always welcome.
>>>
>>> - X3D 4.1 (draft) Architecture, clause 9 Networking component, 9.3.2
>>> X3DUrlObject
>>> - "All *url* fields can hold multiple string values. The strings in
>>> these fields indicate multiple locations to search for data in the order
>>> listed. If the X3D browser cannot locate or interpret the data specified by
>>> the first location, it shall try the second and subsequent locations in
>>> order until a location containing interpretable data is encountered. X3D
>>> browsers only have to interpret a single string. If no interpretable
>>> locations are found, the node type defines the resultant default behaviour."
>>> -
>>> https://www.web3d.org/specifications/X3Dv4Draft/ISO-IEC19775-1v4.1-CD//Part01/components/networking.html#X3DUrlObject
>>> X3D Architecture 4.1 draft — ISO/IEC 19775-1:202x — 9 Networking
>>> component
>>> <https://www.web3d.org/specifications/X3Dv4Draft/ISO-IEC19775-1v4.1-CD//Part01/components/networking.html#X3DUrlObject>
>>> 9.2 Concepts 9.2.1 URLs, URNs and URIs. A URL (Uniform Resource
>>> Locator) is a form of Universal Resource Identifier (URI) described in that
>>> specifies a file located on a particular server and accessed through a
>>> specified protocol (such as file:, http: or https:).In this document, the
>>> upper-case term URL refers to a Uniform Resource Locator, while the
>>> italicized lower-case version url refers to ...
>>> www.web3d.org
>>>
>>>
>>> 3. """Individual font library files can be used by multiple FontStyle
>>> nodes
>>> in a scene.""" -- How? FontStyle does not refer to FontLibrary. And I
>>> understand in this version, FontStyle would not have its own
>>> FontStyle.url
>>> field.
>>>
>>> Did you maybe want to add a field like "SFNode FontLibrary
>>> fontLibrary" to FontStyle or Text? Or make FontLibrary a X3DChildNode?
>>>
>>>
>>> That was my first thought as well but see above.
>>> A simple way to make a FontLibrary url available to multiple FontStyle
>>> nodes is by DEF/USE.
>>> A FontStyle.fontLibrary field is preferred over a global scope since it
>>> clearly associates a font to a FontStyle/Text.
>>> It is better not to have to rely on the family name to identify a font.
>>> Having to rely on the family name is more error prone for both the author
>>> and the browser.
>>>
>>> *response: *the cyan text for FontStyle draft prose shows the
>>> less-desirable alternative. Have added strikethrough (using html <del>
>>> tag) and better description to indicate that we moved url fields from
>>> FontStyle to FontLibrary). The cyan alternative is there now for
>>> comparison, essentially the prose is the same. Hopefully the draft prose
>>> is clearer now, with no duplication of functionality.
>>>
>>> *response: *We are thinking that if inconsistencies exist between a
>>> specified FontStyle family field and the font file, then such cases are
>>> best handled on a best-effort basis by browsers without further complex
>>> (and possibly contradictory) guidance in specification prose. Perhaps this
>>> stance should be a NOTE?
>>>
>>>
>>>
>>>
>>>
>>>
>>> 4. A testcase, preferably many testcase(s), explaining the FontLibrary
>>> usage (X3D files + intended rendering) would very helpful. (Same goes for
>>> the previous proposal actually.) Please provide testcases along with spec
>>> prose -- they make the intention more obvious.
>>>
>>>
>>> Including edge cases if possible.
>>>
>>> * response*: Agreed. We will review the following simple example in
>>> our next editors meeting. More examples are welcome from people
>>> contributing to this list.
>>> EXAMPLE
>>>
>>> FontLibrary { # example derived from TextHaveFunWithX3D.x3d <https://andreasplesch.github.io/Library/Viewer/index.html?url=https://gist.githubusercontent.com/andreasplesch/dc9111dcd106f1a69d567ceca8f52701/raw/8cdf9764f11e08b5f541dee487fc0811bb0cf927/FontHaveFunWithX3D.x3d>
>>> url ["MjQamj1kuP_soQ3o-rysO9Ci_8oJlIUUInI.woff2"
>>> "https://fonts.gstatic.com/s/playwritecaguides/v1/MjQamj1kuP_soQ3o-rysO9Ci_8oJlIUUInI.woff2"]
>>> }
>>> Shape {
>>> # ☺ = ☺ smiley face emoticon, and "" means skip a line
>>> geometry Text {
>>> string [ "Have fun" "with X3D!" "" ":) ☺" ]
>>> fontStyle FontStyle {
>>> spacing 1.5
>>> family=["Playwright_CA_Guides"]
>>> justify [ "MIDDLE" "MIDDLE" ]
>>> }
>>> }
>>> }
>>>
>>>
>>> Overall, so far, I admit FontLibrary is confusing to me, I don't know
>>> what
>>> it means. The previous proposal (extending FontStyle) seems more
>>> straightforward -- a few people (including me) had feedback, but it seems
>>> that we all "got the intention" of FontStyle.url correctly, we just
>>> pointed
>>> some things to improve in wording.
>>>
>>>
>>> One advantage of a FontLibrary (or perhaps just 'Font') node would be
>>> DEF/USE.
>>>
>>> As mentioned multiple times, the spec. should include language that the
>>> FontStyle.style field may be ignored if a FontLibrary url is accessible as
>>> in most cases that url will only provide data for one specific style..
>>>
>>> Regards, -Andreas
>>>
>>> *response: *the presence of the css style field might add a whole new
>>> level of sophistication and possible edge cases.
>>>
>>> This again sounds like a rationale for a NOTE in the specification
>>> specifically pointing out that browsers have latitude to handle ambiguous
>>> cases before falling back to default font... Might be obvious/redundant or
>>> might be helpful.
>>>
>>> Hope these responses help us converge, again thanks for together
>>> thinking things through. v/r Don
>>>
>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://web3d.org/pipermail/x3d-public_web3d.org/attachments/20250304/b0ca0161/attachment-0001.html>
More information about the x3d-public
mailing list