Prefixed S-expressions and Related Markup Techniques #6

AdamSobieski · 2026-08-04T20:00:17Z

AdamSobieski
Aug 4, 2026
Maintainer

Introduction

What if S-expressions could be prefixed with symbols like # and @ to represent attributes and metadata of their parent elements?

Mapping between Prefixed S-expressions and Markup

Using the : symbol as XML does, what if the # symbol indicated that an S-expression was an attribute of its parent expression and the @ symbol indicated that an S-expression was metadata, e.g., an annotation, for its parent expression?

(builtin:resource #(xmlns:builtin "...") #(xmlns:ns "...") #(xmlns:meta "...")
  (ns:element @(meta:data "x") #(ns:attr1 "value") #(ns:attr2 (ns:substructure)) (ns:child))
)

The S-expression, above, could be serialized to XML content resembling:

<builtin:resource xmlns:builtin="..." xmlns:ns="..." xmlns:meta="..." xmlns:x="...">
  <ns:element ns:attr1="value">
    <meta:data x:mode="metadata">x</meta:data>
    <ns:attr2 x:mode="attribute">
      <ns:substructure />
    </ns:attr2>
    <ns:child />
  </ns:element>
</builtin:resource>

Above, the x:mode="attribute" simple attribute and value signals that an element is an extended attribute of its parent element. While simple attributes allow only text-string values, extended attributes allow markup and other content to be used for attributes' values.

Above, the x:mode="metadata" simple attribute and value signals that an element is metadata of its parent element.

Inclusions

Also, drawing inspiration from XML inclusions, perhaps the * symbol could be of use for retrieving referenced S-expression content.

*(x:value #(x:type "text/x-s-expression") #(x:src "https://www.data.org/example/resource.s"))

Markup Examples

Using the above XML serialization for the indicated variety of S-expressions, a number of examples are indicated below, showing the usefulness of the expressiveness.

Outlines

The following examples intend to show that one could readily represent outlines with a format supporting extended attributes and metadata.

Starting with a simple representational skeleton:

<node xmlns="...">
  <node content="Linear electron transport through a photosystem..." />
  <node content="Carbon fixation produces the three-carbon sugar intermediate..." />
</node>

one could add education-related metadata and use extended attributes to allow formats other than text strings for outline nodes' contents:

<node xmlns="..." xmlns:edu="..." xmlns:x="...">
  <edu:objective x:mode="metadata">The objective of this section is to explain photosynthesis.</edu:objective>
  <edu:preconditions x:mode="metadata">...</edu:preconditions>
  <edu:effects x:mode="metadata">...</edu:effects>
  <edu:guiding-questions x:mode="metadata">...</edu:guiding-questions>
  <node>
    <edu:objective x:mode="metadata">The objective of this section is to explain water photolysis.</edu:objective>
    <content x:mode="attribute">
      <x:value x:type="application/xhtml+xml">
        <![CDATA[
          <html xmlns="http://www.w3.org/1999/xhtml"><body>Linear electron transport through a photosystem will leave
          the reaction center of that photosystem oxidized.</body></html>
        ]]>
      </x:value>
    </content>
  </node>
  <node>
    <edu:objective x:mode="metadata">The objective of this section is to explain the Calvin cycle.</edu:objective>
    <content x:mode="attribute">
      <x:value x:type="application/xhtml+xml">
        <![CDATA[
          <html xmlns="http://www.w3.org/1999/xhtml"><body>Carbon fixation produces the three-carbon sugar intermediate,
          which is then converted into the final carbohydrate products.</body></html>
        ]]>
      </x:value>
    </content>
  </node>
</node>

Above, the special element <x:value> signals that a metadata or an attribute has a value of a type other than default text/plain text strings.

Workflows

The following example intends to show that one could readily represent workflows with a format supporting extended attributes and metadata.

<sequence xmlns="..." xmlns:x="...">
  <code name="start-activity" xmlns="...">
    <execute x:mode="attribute">
      <x:value x:type="text/javascript" x:src="script.js" x:fragment="onStart" />
    </execute>
  </code>
  <code name="process-activity" xmlns="...">
    <execute x:mode="attribute">
      <x:value x:type="text/javascript" x:fragment="OnProcess">
        <![CDATA[
           function onProcess(ctx)
           {
              ...
           }
        ]]>
      </x:value>
    </execute>
  </code>
  <code name="end-activity" xmlns="...">
    <execute x:mode="attribute">
      <x:value x:type="text/javascript" x:src="script.js" x:fragment="onEnd" />
    </execute>
  </code>
</sequence>

Abstract Syntax Trees

The following example intends to show that one could readily represent abstract syntax trees with a format supporting extended attributes and metadata.

<query xmlns="..." xmlns:x="...">
  <select>
    <selector x:mode="attribute">
      <x:value x:type="application/jsast+xml;version=es2026">
        <![CDATA[
          <js:lambda xmlns:js="..." xmlns:x="...">
            <js:parameters x:mode="attribute">
              <js:parameter js:name="r" />
            </js:parameters>
            <js:body x:mode="attribute">
              <!-- ... -->
            </js:body>
          </js:lambda>
        ]]>
      </x:value>
    </selector>
    <where>
      <functor x:mode="attribute">
        <x:value x:type="application/jsast+xml;version=es2026">
          <![CDATA[
            <js:lambda xmlns:js="..." xmlns:x="...">
              <js:parameters x:mode="attribute">
                <js:parameter js:name="r" />
              </js:parameters>
              <js:body x:mode="attribute">
                <!-- ... -->
              </js:body>
            </js:lambda>
          ]]>
        </x:value>
      </functor>
      <from>
        <src x:mode="attribute">
          <x:value x:type="application/csvm+json" x:src="data.json" />
        </src>
      </from>
    </where>
  </select>
</query>

Multipart Data

See also: #5

The <x:value> special element, having a x:src attribute, is well-suited for utilization within multipart MIME messages. Using RFC 2392's mid: and cid: URL schemes, data could be stored as other message parts and referenced by URL in messages' parts.

Click here to toggle view of a multipart MIME message containing a part using these markup features.

Mime-Version: 1.0
Message-ID: <12345678>
Content-Type: multipart/related; boundary="boundary-example"

--boundary-example

Content-ID: <part1>
Content-Type: application/xml

<node xmlns="..." xmlns:edu="..." xmlns:x="...">
  <edu:objective x:mode="metadata">The objective of this section is to explain photosynthesis.</edu:objective>
  <edu:preconditions x:mode="metadata">...</edu:preconditions>
  <edu:effects x:mode="metadata">...</edu:effects>
  <edu:guiding-questions x:mode="metadata">...</edu:guiding-questions>
  <node>
    <edu:objective x:mode="metadata">The objective of this section is to explain water photolysis.</edu:objective>
    <content x:mode="attribute">
      <x:value x:src="cid:part2" />
    </content>
  </node>
  <node>
    <edu:objective x:mode="metadata">The objective of this section is to explain the Calvin cycle.</edu:objective>
    <content x:mode="attribute">
      <x:value x:src="cid:part3" />
    </content>
  </node>
</node>

--boundary-example

Content-ID: <part2>
Content-Type: application/xhtml+xml

<html xmlns="http://www.w3.org/1999/xhtml"><body>Linear electron transport through a photosystem will leave
the reaction center of that photosystem oxidized.</body></html>

--boundary-example

Content-ID: <part3>
Content-Type: application/xhtml+xml

<html xmlns="http://www.w3.org/1999/xhtml"><body>Carbon fixation produces the three-carbon sugar intermediate,
which is then converted into the final carbohydrate products.</body></html>

--boundary-example--

The <x:value> special element could, itself, be used to represent multipart data for content-negotiation scenarios, internationalization and localization scenarios, and other use-case scenarios.

<resource xmlns="..." xmlns:x="...">
  <content x:mode="attribute">
    <x:value x:type="multipart/alternative">
      <x:value x:type="text/plain" x:lang="en">
        <![CDATA[
          Carbon fixation produces the three-carbon sugar intermediate,
          which is then converted into the final carbohydrate products.
        ]]>
      </x:value>
      <x:value x:type="text/plain" x:lang="fr">
        <![CDATA[
          La fixation du carbone produit un intermédiaire glucidique à trois
          carbones, qui est ensuite transformé en produits glucidiques finaux.
        ]]>
      </x:value>
    </x:value>
  </content>
</resource>

The following example shows a markup representation of some preliminary "MIME-LD" concepts, explored in greater detail in #5 .

<resource xmlns="..." xmlns:x="...">
  <content x:mode="attribute">
    <x:value x:type="multipart/related" x:mid="12345678" x:context="https://www.example.org/structure.mimeld">
      <x:value x:type="text/turtle" x:cid="part1" x:rel="metadata">
        <![CDATA[
          @prefix dc: <http://purl.org/dc/terms/> .
          @prefix example: <http://example.org/> .

          <mid:12345678> dc:hasPart <cid:part1>, <cid:part2>, <cid:part3> ;
              example:hasMainPart <cid:part2> .

          <cid:part1> dc:format "text/turtle" .

          <cid:part2> dc:format "text/plain" ;
              dc:language "en" ;
              dc:references <cid:part3#archer2010climate> .

          <cid:part3> dc:format "text/x-bibtex" .
        ]]>
      </x:value>
      <x:value x:type="text/plain" x:lang="en" x:cid="part2" x:rel="content">
        <![CDATA[
          Climate change is caused by human activities. Therefore, we should reduce carbon emissions.
        ]]>
      </x:value>
      <x:value x:type="text/x-bibtex" x:cid="part3" x:rel="supplemental">
        <![CDATA[
          @book{archer2010climate,
              title={The climate crisis: An introductory guide to climate change},
              author={Archer, David and Rahmstorf, Stefan},
              year={2010},
              publisher={Cambridge University Press}
          }
        ]]>        
      </x:value>
    </x:value>
  </content>
</resource>

Extensibility and MIME Headers

Drawing inspiration from the HTTP Vocabulary in RDF, the expressiveness to make use of an extensible, open-ended set of MIME headers could be provided for the <x:value> special element.

<x:value x:type="application/vnd.openxmlformats-officedocument.spreadsheetml.sheet">
  <x:headers x:mode="attribute">
    <x:header x:fieldName="Content-Description" x:fieldValue="This is a description of the message part." />
  </x:headers>
  <![CDATA[
    ...
  ]]>
</x:value>

Artificial Intelligence Multimodal Chat Messages and Histories

See also: vCon, ChatMessageContent, ChatHistory

Coming soon.

Conclusion

What do you think about these ideas? Thank you.

Navigation Menu

Prefixed S-expressions and Related Markup Techniques #6

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{editor}}'s edit

{{editor}}'s edit

Uh oh!

Replies: 0 comments

Select a reply

Uh oh!

Prefixed S-expressions and Related Markup Techniques #6

Uh oh!

Uh oh!

AdamSobieski Aug 4, 2026 Maintainer

Introduction

Mapping between Prefixed S-expressions and Markup

Inclusions

Markup Examples

Outlines

Workflows

Abstract Syntax Trees

Multipart Data

Extensibility and MIME Headers

Artificial Intelligence Multimodal Chat Messages and Histories

Conclusion

Replies: 0 comments

AdamSobieski
Aug 4, 2026
Maintainer