[x3d-public] Problem in x3dviewscene: ROUTE placement

Mon Jun 12 06:54:35 PDT 2023

Thank you for identifying the relevant seection of the X3D V4 Architecture. The relevant sentence from section 4.4.8.2 is:

ROUTE statements may either appear at the top level of an X3D file or inside a node wherever fields may appear.

It declares the intent of allowing a ROUTE statement inside a node definition in close proximity to a field that the ROUTE modifies, for purpose of readability, but other than that purpose  this placement of the ROUTE statement does not change the semantics of a ROUTE statement.  I don't think this sentence bears being taken literally to an extreme; otherwise we are obligated to define a additional XML encoding of the ROUTE statement which allows the ROUTE statement to be included as an XML attribute; so that, for example, it can be placed adjacent to the XML encoding of the SFVec3f "translation" field of the Transform element. I am not trying to argue by exaggeration;t I am argujng that the sentence need to be interpreted in accordance with the syntax of a specific encoding.

The ClassicVRML encoding does allow a ROUTE to be defined in the body of a X3D node; the specific issue we are discussing is whether this includes placing the ROUTE definition inside the ClassicVRML encoding of an MFNode value. We could have a continued discussion about whether inclusion as an item in a MFNode value is included in the text phrase "..wherever fields may appear", but we can also refer to the formal grammar definition of the ClassicVRML appendix which, in my judgement, does not allow that.

Taking a step back, and asuming that the JoeLevel2LOA3SSPYRWRJKHud.x3dv example was generated by a tool; we have one tool that generates an x3dv file which does not conform to the ClassicVRML formal geometry published in X3D V3.3 . If I understand Don's email report of June 9, the X3D conversion of the x3dv file passed validation, but this was not  direct validation of the original x3dv file. So I see the validation situation as:

1) The view3dscene code identifies the ClassicVRML error in JoeLevel2LOA3SSPYRWRJKHud.x3dv when asked to render it,'
2) The conversion code, from x3dv -> x3d, offered by view3dscene, does not flag an error in JoeLevel2LOA3SSPYRWRJKHud.x3dv but does generate a valid x3d scene from it,

In the example archives, we have not yet identified a ClassicVRML encoding that includes this very specific feature, a ROUTE statement included within an MFNode value of an X3D field, so it is not yet possible to use our example archive for guidance.

I think these considerations make it premature to change the ClassicVRML grammar definition in the standard.

Vince Marchetti

> On Jun 12, 2023, at 1:36 AM, Brutzman, Donald (Don) (CIV) <brutzman at nps.edu> wrote:
> 
> 	• Sorry to be unclear, that was not the relevant reference, rather the prior email in the thread.
>  
> My mail of 9 June 16:30:04 PDT 2023 holds the relevant reference.  Excerpt follows, quoting a quote, and have provided yellow highlights to key phrasing this time.
>  
>  
> 	• [x3d-public] Problem in x3dviewscene: ROUTE placement
> 	• http://web3d.org/pipermail/x3d-public_web3d.org/2023-June/018836.html
>  
> The intent for ROUTE is that it might appear within other nodes.  The
> phrasing in X3D4 Architecture is quite explicit about this:
>  
> * X3D 4.0 Part 1: Architecture and base components, clause 4 Concepts, 4.4.8.2 Routes
> * https://www.web3d.org/specifications/X3Dv4Draft/ISO-IEC19775-1v4-IS.proof/Part01/concepts.html#Routes
>  
> *       Routes allows an author to declaratively connect the output events
> of a node to input events of other nodes, providing a way to implement
> complex behaviors without imperative programming. When a routed output event
> is fired, the corresponding destination input event receives notification
> and can process a response to that change. This processing can change the
> state of the node, generate additional events, or change the structure of
> the scene graph. Routes may be created declaratively in an X3D file or
> programmatically via an SAI call.
> *       Routes are not nodes. The ROUTE statement is a construct for
> establishing event paths between specified fields of nodes. ROUTE statements
> may either appear at the top level of an X3D file or inside a node wherever
> fields may appear. A ROUTE statement shall only appear after the definition
> of the source and destination nodes. Placing a ROUTE statement within a node
> does not associate it with that node in any way. A ROUTE statement does
> follow the name scoping rules as described in 4.4.7 Run-time name scope.
>  
>  
> 	• I next took the offered example scene
> 	• https://www.web3d.org/x3d/content/examples/Vrml2Sourcebook/Chapter08AnimatingPositionOrientationScale/Figure08_3PositionInterpolator.x3d
>  
> and moved entire scene graph within Scene inside a Group, i.e.
> 	• <Scene><Group><!—everything including ROUTE statements --></Group></Scene>
> 	• Thus nothing remains in top-level MFNode list of the scene, except for single parent Group
>  
> Then tested the modified example.
> 	• All validation tests passed.
> 	• Xj3D worked.
> 	• X3DOM worked.
> 	• After 10 minutes of reporting to Microsoft that view3dscene was legitimate, and coercing Windows 10 to accept it, view3dscene.exe worked on this modified example too.
> 	• Searching for “<ROUTE” found 4989 matches in first 500 scenes searched (out of 4060 total archived scenes) so no doubt there are more.
>  
> 	• To my prior list of benefits for flexible placement of ROUTE statements, here is another:
> 	• Simplifies author and tool addition of code-block patterns and templates within a scene (this a feature used numerous times as an optional authoring assist for event tracing by X3D-Edit, likely to be added as utility features in X3DJSAIL and X3DPSAIL someday as well.)
> Specifically, from X3D-Edit documentation,
> 	• Trace capabilities are available for ROUTE, sensors, interpolators and other nodes that produce or consume events. Trace output results are logged on the browser console to show the timing and values passed events. When the Trace checkbox is selected by the author, a Script node is inserted immediately after the node to accomplish this task. This is a useful debugging technique.
> 	• https://savage.nps.edu/X3D-Edit/images/TraceFeatureReportsRoutedEventValuesOnBrowserConsoleViaScriptOutput.png
> 	• https://x3dgraphics.com/examples/X3dForWebAuthors/Chapter07-EventAnimationInterpolation/Chapter07-EventAnimationInterpolation-EventTracing.pdf
>  
> 	• “Consistency across all file encodings and language bindings is possible.” This is our basic design principle for separation of functionality in X3D architecture from implementation details in all related specifications.
>  
> We have immense interoperability already.  As promised many times, we will smooth out any mismatches in future X3D 4.0 file encodings and language bindings.  Test cases do help too.  So stand by for action.
>  
> Hopefully the highlighted specification requirement, the ease of creating flexible test cases, and the value of this existing specification capability is presented more clearly now.
>  
> Careful consideration, implementation efforts, and consensus-based progress are always appreciated.
>  
> Having fun with X3D… I hope that others are too.
>  
> 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: vmarchetti at kshell.com <vmarchetti at kshell.com> 
> Sent: Sunday, June 11, 2023 4:42 PM
> To: Brutzman, Donald (Don) (CIV) <brutzman at nps.edu>; X3D-Public <x3d-public at web3d.org>
> Cc: Michalis Kamburelis <michalis.kambi at gmail.com>
> Subject: Re: [x3d-public] Problem in x3dviewscene: ROUTE placement
> 
> 
> On Jun 11, 2023, at 5:09 PM, Brutzman, Donald (Don) (CIV) <brutzman at nps.edu> wrote:
>  
> Not trying to invent anything here, rather point out what is specified (and also common practice).  I already gave the relevant links in the X3D 4.0 Architecture governing what is required.  
>  
> I do not see this requirement given or even implied in the link you specified:
> 	• https://www.web3d.org/specifications/X3Dv4Draft/ISO-IEC19775-1v4-IS.proof/Part01/fieldTypes.html
>  
> is a comprehensive listing of the X3D Field types, with details on SFNode and MFNode therein, and on this page I don't see any mention of the ROUTE statement nor does a text search on that page discover even the terms "route" or "statement" in isolation.
>  
> ...
> 
> 
> Consistency across all file encodings and language bindings is possible.
> I don't this this statement is true, or to the extent that it is true, then long - ago X3D design decisions did not treat it as of primary importance. As Michalis pointed out, in the ClassicVRML encoding the zero or more items in a single MFNode value are syntactically enclosed in square brackets. Moreover, in the ClassicVRML encoding all the field values are explicitly identified with the name of the field. However, in the X3D encoding the XML child elements are in most cases identified as belonging to an X3D field through the node type-inheritance classification; and in the few cases where that is not possible an explicit "containerField" XML attribute is used. To make the X3D encoding consistent with ClassicVRML would entail the use of wrapper XML elements around the element encoding of one node-element for an SFNode value,  or the several nodes of MFNode. However, the design decision was made long ago not to use wrapper elements in the XML encoding ( see https://www.web3d.org/x3d/content/examples/Basic/development/WrapperTagsConsideredHarmful.html )
>  
> My point is that the consistency among encodings is a useful criteria but should not automatically take precedence over other considerations. One such consideration is maintainging stability of our formal ClassicVRML grammar.
>  
> And I think there is little additional benefit in changing the the ClassicVRML to allow ROUTE inside MFNode encodings. Doing so does allow a ROUTE to be placed adjacent to a node whose fields it modifies and you (Don) have pointed out the readability benefits that offers -- but Michalis points out that the ClassicVRML already allows that ROUTE statement to be placed inside the body of a node it modifies, even closer!
>  
> Vince Marchetti