[x3d-public] C/C++/C#---X3DCSAIL; specification and source meeting minutes

[John Carlson]

Myeong, thank you for your work on X3DCSAIL.

There are currently 2 styles of library generation, xslt stylesheet and
python reading X3DUOM.   The python approach breaks down to whole library
generation and class reference implementation—“auto class” built on top of
X3DJSAIL.   I am currently working on an ECMAScript 6 library without auto
class.

You may choose to do the proven xslt stylesheet method (declarative,
proven, supported by Don and NPS).  I have chosen python to avoid writing
in XML.

Generally speaking, I would prefer a library compiled by emscripten and
LLVM, such that it could be deployed to WebAssembly and WASI.  Perhaps
using empscripten on your existing source code would be a good next thing
to do. If we have WebAssembly as a base, that invites programming languages
to use our libraries, and supports the whole X3DV4 ideas for web
integration.   We already have JavaScript and Python solutions on the web,
why not C/C++/C#?

If we could convert xslt to WebAssembly, that would be very cool.   It’s
probably not our bailiwick—I’m just trying to lure Don into the web.


Yes, we should look at performance between WebAssembly and non-WebAssembly
as well.

If you like, I will try to accomplish the compilation to WebAssembly task.

Thanks!

John

On Fri, Sep 4, 2020 at 10:14 PM Myeong Won Lee <myeongwonlee at gmail.com>
wrote:

> Dear John,
>
> Thank you for your email.
>
> We didn't use autogeneration, and all the classes and functions in Annex A
> and B in the text of the specification were rewritten manually.
> In other words, with one by one editing.
> A huge amount of time was taken to write them.
>
> In addition, the X3DLib has partial libraries (binary) implemented among
> them.
> Partial X3D components have been implemented.
> We will enhance the X3DLib by incorporating the classes and functions in
> Annex A and B.
>
> We must learn how to use autogeneration for the code.
> After getting the results of autogeneration, we will figure out the
> performance gain of auto generation by comparing our manually generated
> code and the autogenerated code.
>
> Sincerely,
>
> Myeong
>
> On Fri, Sep 4, 2020 at 10:56 AM John Carlson <yottzumm at gmail.com> wrote:
>
>> What we need is an easy way to upgrade the libraries using the X3DUOM
>> when the standard changes.  If you have a tool which does that, that would
>> be great!  Otherwise, your example is great, and we will pursue
>> autogeneration of the libraries as a separate activity, following your
>> pattern.
>>
>> Does that make sense?
>>
>> John
>>
>> On Thu, Sep 3, 2020 at 6:09 PM Myeong Won Lee <myeongwonlee at gmail.com>
>> wrote:
>>
>>> Dear Don,
>>>
>>> In the X3D C/C++ implementation code, the reason that there are no
>>> sources for X3D nodes is that the X3D viewer reads and parses the example
>>> X3D file and renders a X3D Box using the X3DLib directly.
>>> Does this make sense to you?
>>> The X3DLib includes a huge number of functions as a library.
>>> Because it seems necessary to include the C++ source code for the
>>> example X3D nodes in Annex C, we will provide them separately from X3DLib.
>>>
>>> Sincerely,
>>>
>>> Myeong
>>>
>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> ------------ 원본 메일 내용 ------------
>>>> *보낸 사람 :* "Don Brutzman"<brutzman at nps.edu>
>>>> *받는 사람 :* "X3D Graphics public mailing list" <x3d-public at web3d.org>
>>>> *참      조 :* "John Carlson" <yottzumm at gmail.com>, "Roy Walmsley" <
>>>> roy.walmsley at ntlworld.com>, "Myeong Won Lee" <mwlee at suwon.ac.kr>,
>>>> "Doug Sanden" <gpugroup at gmail.com>
>>>> *메일 제목 :* Re: C/C++/C#---X3DCSAIL; specification and source meeting
>>>> minutes
>>>> *보낸 날짜 :* Wed, 2 Sep 2020 20:09:42 -0700
>>>>
>>>>
>>>>
>>>>
>>>> Attendees: MWL, DPB
>>>>
>>>> We are finally back in motion on these specifications and met today,
>>>> making progress.
>>>>
>>>> Full design plan (produced earlier) appears below today's minutes.
>>>>
>>>> ---
>>>>
>>>> 1. Specifications.
>>>>
>>>> The Github site is where Web3D Consortium members have full access to
>>>> draft specifications.
>>>>
>>>> [1.1] Github, Web3D Consortium, X3D
>>>> https://github.com/Web3DConsortium
>>>> https://github.com/Web3DConsortium/X3D
>>>>
>>>> [1.2] X3D SAI Language Bindings
>>>> https://github.com/Web3DConsortium/X3D/tree/master/ISO-IEC19777
>>>>
>>>> [1.3] X3D 19777-3 C Language Bindings
>>>>
>>>> https://github.com/Web3DConsortium/X3D/tree/master/ISO-IEC19777/ISO-IEC19777-3/ISO-IEC19777-3v3.3/ISO-IEC19777-3v3.3-CD
>>>>
>>>> [1.4] X3D 19777-4 C++ Language Bindings
>>>>
>>>> https://github.com/Web3DConsortium/X3D/tree/master/ISO-IEC19777/ISO-IEC19777-4/ISO-IEC19777-4v3.3/ISO-IEC19777-4v3.3-CD
>>>>
>>>> [1.5] X3D 19777-5 C# Language Bindings
>>>>
>>>> https://github.com/Web3DConsortium/X3D/tree/master/ISO-IEC19777/ISO-IEC19777-5/ISO-IEC19777-5v3.3/ISO-IEC19777-5v3.3-CD
>>>>
>>>> [1.6] X3D 19777 New Work Item Proposals (NWIP)
>>>> https://github.com/Web3DConsortium/X3D/tree/master/ISO-IEC19777/NWIP
>>>>
>>>> Looking at each of the above Committee Draft (CD) directories, they
>>>> were last touched 8 March 2019 (18 months ago).
>>>>
>>>> *Myeong has updates to each of these. She is sending zip files and I
>>>> will commit them into github.*
>>>>
>>>> Because they have changed, and because we are now working to meet ISO
>>>> deadlines, we think that these should all be saved separately as an updated
>>>> Working Draft #1. So new directories will say "WD1" instead of "CD" to
>>>> match our naming conventions.
>>>>
>>>> Our next question is whether we should go to version 4.0 (rather than
>>>> 3.3). This seems like a good idea... but there are several issues.
>>>>
>>>> a. Are they complete for 4.0?
>>>> b. Does the relevant SAI 19775-2 for X3D also have to be updated?
>>>> c. Can we autogenerate the new parts of these language bindings to
>>>> match X3D4.
>>>>
>>>> We? are hoping that, following some further editorial work, that the
>>>> answer to all three of these questions can become "yes" for X3D version
>>>> 4.0.
>>>>
>>>> [1.7] Web3D Recommended Standards
>>>> https://www.web3d.org/standards
>>>>
>>>> [1.8] X3D 19775-2, ISO/IEC 19775-2:2015, V3.3,
>>>> X3D Abstract : Scene Access Interface (SAI)
>>>> International Standard (IS), 2015-04-24
>>>> https://www.web3d.org/documents/specifications/19775-2/V3.3/index.html
>>>>
>>>> A quick review of Part 2: Scene access interface (SAI) looks like very
>>>> very few changes will occur for the transition from X3D Architecture v3.3
>>>> to v4.0.
>>>>
>>>> Recommended next steps:
>>>> d. Have a dedicated meeting of X3D Working to review both specification
>>>> and Mantis issues list to see what changes are needed to create 19775-2
>>>> version 4.0.
>>>> e. If indeed small, then we create such a working draft on github.
>>>> f. If large, then we will defer to January 2021 following expected
>>>> completion of X3D 19775-1 X3D version 4.0
>>>> g. We work on version 4 for all language bindings so that we don't have
>>>> to repeat further specification revisions later.
>>>>
>>>> Definitely the sequencing of this work needs to be considered in light
>>>> of ISO deadlines and our ability to accomplish work.
>>>>
>>>> Rephrase: if it is possible to go fast, let's try. Parallel work is
>>>> likely possible since functionality is stable. If we are too busy, then we
>>>> wait for January.
>>>>
>>>> Dick please review and verify how we might best handle these
>>>> specification drafts for ISO.
>>>>
>>>> ---
>>>>
>>>> 2. Example implementations.
>>>>
>>>> Suwon University provided some example source code last year.
>>>>
>>>> Two types of code are possible. First type is simply for handling data
>>>> structures; second type is a renderer.
>>>>
>>>> Last year's examples are all sitting in Web3D version control on
>>>> SourceForge, with 5 subdiretories:
>>>>
>>>> [2.0] X3D Sourceforge subversion arcchives
>>>>
>>>> https://sourceforge.net/p/x3d/code/HEAD/tree/www.web3d.org/x3d/languages/
>>>> =
>>>>
>>>> a. c, not seen: source code for nodes, or presentation on this work.
>>>> b. cpp, not seen: source code for nodes, or presentation on this work.
>>>> c. csharp, found: both source code for nodes and README presentation
>>>>
>>>> Myeong has an excellent long presentation that describes many things in
>>>> these archives, she will send and we will add.
>>>>
>>>> At first is seems like the c and cpp directories do not include source
>>>> implementations for each and every node (Box, Group, Appearance etc.)
>>>> because they are actually found in the X3DLib .h files. There is typically
>>>> a mix of Abstracts.h Concretes.h and X3DLib.h files, be sure to look
>>>> closely. For example code patterns, see the following:
>>>>
>>>> d. c,
>>>> https://sourceforge.net/p/x3d/code/HEAD/tree/www.web3d.org/x3d/languages/c/X3DLib/Concretes.h
>>>> e. cpp,
>>>> https://sourceforge.net/p/x3d/code/HEAD/tree/www.web3d.org/x3d/languages/cpp/X3DLib/Concretes.h
>>>> f. csharp,
>>>> https://sourceforge.net/p/x3d/code/HEAD/tree/www.web3d.org/x3d/languages/csharp/3.X3DLib/X3DNode/
>>>> (many files)
>>>>
>>>> Although all of this code was created manually (OMG!) everything looks
>>>> to me like well-defined code patterns that might autogenerated from X3DUOM.
>>>>
>>>> d. python: contains X3dPythonViewer which is a pure python viewer for
>>>> X3D. This uses manual patterns but might be upgraded (by someone brave and
>>>> magnificent!) to match our upgraded code patterns for X3DPSAIL.
>>>>
>>>> [2.1] Python X3D Package x3d.py
>>>> https://pypi.org/project/x3d
>>>>
>>>> [2.2] X3D Python Scene Access Interface Library (X3DPSAIL)
>>>> https://www.web3d.org/x3d/stylesheets/python/python.html
>>>>
>>>> e. suwon (temporary holding directory, used only for sharing files
>>>> prior to production)
>>>>
>>>>
>>>> To use X3DUOM to create both codebases and draft specification annexes,
>>>> we can follow the existing solutions already provided for Java and Python.
>>>>
>>>> Given today's code review, I can confirm that we can follow the exact
>>>> approaches used already for Java and Python. The following summary provide
>>>> much detail.
>>>>
>>>> ---
>>>>
>>>> 3. Next steps.
>>>>
>>>> a. Specification: Myeong Dick and I can work on the sources.
>>>>
>>>> b. Example implementations: *who wants to help* with the C, C++, C#
>>>> interfaces? We need expert review on these interfaces and sample
>>>> implementations before we try to standardize them.
>>>>
>>>> c. Working group meeting: review scope of work to upgrade 19775-2 SAI
>>>> to X3D 4.0. Soon, within a month.
>>>>
>>>> d. Specification editors meeting, Myeong Don Dick: latest versions of
>>>> documents in github. We will try to meet next week if that works for Dick.
>>>>
>>>> e. Codebase generation meeting: when one or more volunteers are ready
>>>> to help with review, we can then discuss.
>>>>
>>>> Have fun with X3D using C, C++, C# !! 8)
>>>>
>>>> Yours truly, Myeong and Don
>>>>
>>>>
>>>> On 8/11/2020 7:01 AM, Don Brutzman wrote:
>>>> > John, thanks for posting.  There are a number of steps still to go,
>>>> here is more information.
>>>> >
>>>> > a. We currently 2/3 complete in our virtual 3-week-long annual
>>>> meeting for ISO Standards Committee 24.  Ordinarily this is a dedicated
>>>> 1-week face-to-face meeting, it is quite involved.
>>>> >
>>>> > b. Dr. Myeong Won Lee of Suwon University has provided an update to
>>>> the C/C++/C# language bindings for X3D that were first proposed two years
>>>> ago.
>>>> >
>>>> > [1]    Web3D Consortium Members are now considering C, C++ and C#
>>>> language bindings for
>>>> >     #X3D Graphics International Standard. Member value continues to
>>>> grow, early adopters are advised to join!
>>>> >     All community feedback is welcome too.
>>>> >     https://twitter.com/Web3DConsortium/status/984257256270135296
>>>> >
>>>> > c. The Scene Access Interface (SAI) language bindings are strictly
>>>> defined by X3D specifications.  No proprietary technology or
>>>> external-library dependencies are required.
>>>> >
>>>> > [2]    Web3D Recommended Standards
>>>> >     https://www.web3d.org/standards
>>>> >
>>>> > d. We have captured all X3D scene-graph structure and typing in the
>>>> X3D Unified Object Model (X3DUOM).
>>>> >
>>>> > [3]    X3D Unified Object Model (X3DUOM)
>>>> >     https://www.web3d.org/specifications/X3DUOM.html
>>>> >
>>>> > e. We have achieved autogeneration from X3DUOM to open-source
>>>> implementations that support SAI for Java, Python and Turtle.  This occurs
>>>> through careful consideration of design patterns to hold strictly designed
>>>> data structures.  This does not include any form of rendering or
>>>> platform-specific adaptations.
>>>> >
>>>> > [4]    X3D Graphics Standards Relationships
>>>> >
>>>> https://www.web3d.org/specifications/X3dSpecificationRelationships.png
>>>> >
>>>> > f. A major benefit of autogeneration is that everything is exactly
>>>> correct and can always track X3D4 refinements.
>>>> >
>>>> > g. As you know, we hope to repeat this pattern for JavaScript.  Your
>>>> experience with X3DJSONLD and JavaScript means that an autogenerated
>>>> implementation oriented to Node.js is appealing.
>>>> >
>>>> > h. Once implementations are stable, we will have to write
>>>> corresponding X3D SAI specifications to match, likely next spring.
>>>> >
>>>> > i. Professor Lee has also produced an example code library that
>>>> should serve as a good basis for design templates.
>>>> >
>>>> > j. We hope to begin tackling this new task after SIGGRAPH.
>>>> >
>>>> > k. Since an unavoidable FAQ will be "How are the C/C++/C#
>>>> implementations and specifications similar, and how are they different?"
>>>> the X3D working group currently plans to work all three in tandem and
>>>> submit all three specifications simultaneously.  This approach will
>>>> maximize consistency and best-of-breed design patterns.
>>>> >
>>>> > l. Clearly we will need multiple X3D experts who are each familiar
>>>> with C, C++ and C# in order to achieve our goals of useful reusable
>>>> open-source libraries and specifications.
>>>> >
>>>> > m. This is not a debate about programming styles or preferences,
>>>> rather how do we produce the most platform-neutral APIs that are widely
>>>> reusable.
>>>> >
>>>> > n. Resulting codebases will be hooked up to regression testing of the
>>>> X3D Examples Archives.
>>>> >
>>>> > [5]    X3D Resources, Examples: Scene Archives for X3D
>>>> >
>>>> https://www.web3d.org/x3d/content/examples/X3dResources.html#Examples
>>>> >
>>>> > o. Success is measured by how these libraries encourage adoption of
>>>> X3D by C/C++/C# applications, either open source or commercial. Original
>>>> announcement follows.
>>>> >
>>>> > [6]    [x3d] X3D C and C++ language bindings NWIPs [corrected copy]
>>>> >
>>>> http://web3d.org/pipermail/x3d-public_web3d.org/2018-April/008598.html
>>>> >
>>>> > p. I'll be sending out an annual Web3D ISO liaison report to Web3D
>>>> Consortium members in another week, once the SC24 meeting is complete.
>>>> This report will get released publicly sometime afterwards.
>>>> >
>>>> > q. So the answer to most of your questions below is "same as before"
>>>> with Java, Python, Turtle.
>>>> >
>>>> > Hope this makes sense.  These are new big projects, adapting and
>>>> repeating capabilities that we have already demonstrated three times.
>>>> >
>>>> > Worthy mountains to climb that make X3D ever-more available.  Usually
>>>> this is where business plans add a sentence saying, approximately, "total
>>>> world domination to follow."  8)
>>>> >
>>>> > Thanks everyone for considering the possibilities.  Have fun with
>>>> X3D4!  8)
>>>> >
>>>> >
>>>> > On 8/10/2020 5:10 PM, John Carlson wrote:
>>>> >>
>>>> >> We're looking for some programmers to contribute to a project
>>>> creating X3DCSAIL, a C/C++/C# version of X3DJSAIL.
>>>> >>
>>>> >> My question is, how will X3DCSAIL be different from OpenInventor or
>>>> Coin3D or FreeWRL?  Can these run headless?   Can we adapt one of these to
>>>> run headless?  It looks like OpenInventor has a headless mode, but I'm not
>>>> sure about the others.
>>>> >>
>>>> >> How might we pursue this library?  What are the requirements?  Who
>>>> are the customers?   Should we just create examples, and adopt one of the
>>>> existing open source X3D browser implementations?  Is there a website
>>>> describing the goals of this project?  Should I create one?
>>>> >>
>>>> >> I have something written in C++ that reads in X3D JSON and writes
>>>> out X3D XML, but that's pretty much a prototype at this stage--works with a
>>>> couple of cases.  This opens up C++ X3D browser to the X3D JSON
>>>> capabilities similar to how X3DJSONLD transformed X_ITE and X3DOM.
>>>> >>
>>>> >> The cplusplus version in in <X3DJSONLD>/src/main/cplusplus/src
>>>> >>
>>>> >> More testing is welcome!
>>>> >>
>>>> >> We currently do not have a X3dToC.xslt, but would appreciate
>>>> examples of C++ code that people would like to develop to use X3DCSAIL, so
>>>> we can pattern the appropriate stylesheet for the X3DCSAIL library.
>>>> >>
>>>> >> O can provide a C++ Serializer that will convert a DOM document to
>>>> C++ using JavaScript code, but  I'm also wondering what kind of target I
>>>> should hit, similar to what we might do with a stylesheet.
>>>> >>
>>>> >> So we're just looking for C++ test cases at this point.
>>>>
>>>> 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
>>>> http://faculty.nps.edu/brutzman
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>
>>> --
>>>
>>> Myeong Won Lee, PhD, Professor
>>>
>>>
>>> Faculty of Computer Science, U. of Suwon
>>>
>>>
>>> Hwaseong, Gyeonggi-do, 18323 Korea
>>> E-mail) myeongwonlee at gmail.com, mwlee at suwon.ac.kr
>>>
>>>
>>>
>>
>>
>
> --
>
> Myeong Won Lee, PhD, Professor
>
>
> Faculty of Computer Science, U. of Suwon
>
>
> Hwaseong, Gyeonggi-do, 18323 Korea
> E-mail) myeongwonlee at gmail.com, mwlee at suwon.ac.kr
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://web3d.org/pipermail/x3d-public_web3d.org/attachments/20200904/70d0aed1/attachment-0001.html>