Adds explicit definition of the algebraic syntax #227

hartig · 2025-05-25T19:10:26Z

This PR is the next step towards addressing the issue that the spec represents the operators of the algebraic syntax of SPARQL in exactly the same way as it represents the algebra functions used to define the evaluation semantics. See the previous PR for a description of the issue: #216

One aspect of the issue is that the algebraic syntax is not actually defined explicitly. Instead, the possible expressions of the algebraic syntax are introduced only implicitly as part of the translation procedure defined in Section 18.2 Translation to the SPARQL Algebra.

This PR focuses on that aspect by adding a new section that provides an explicit definition of the algebraic syntax, added directly before Section 18.2 Translation to the SPARQL Algebra. This definition includes id anchors for all operators of the algebraic syntax. These anchors can be used in the next step (future PR) to link all mentions of these operators to their definition, similar to how PR #216 linked the algebra operators to their definitions. In fact, as a second part of this PR, Sections 18.2.2.3 Translate Property Path Expressions and 18.2.2.4 Translate Property Path Patterns are already changed to use such links.

I am leaving it to a future PR to do the same (i.e., using such links) everywhere else in Section 18.2 Translation to the SPARQL Algebra and also in Section 18.5.2 Evaluation Semantics. I chose not to include all these changes already in this PR to make it easier to review this step first.
(Section 18.3 after this PR)

Preview | Diff

…he translation of PP expressions/patterns

Tpt

Thank you for pushing this!

Tpt · 2025-05-25T19:26:23Z

spec/index.html

+            <a href="#defn_absOrderBy" class="absOp">OrderBy</a>(|A|, |condition|)
+            is an algebraic query expression if
+            |A| is an algebraic query expression and
+            |condition| is an ordering condition.</li>


It might be nice to link here to a definition of "ordering condition"

I totally agree!!! Yet, there is no such definition anywhere in the spec; at least, I couldn't find any.

The existing parts related to the definition of ORDER BY have the same issue:

Section 18.2.5.1 ORDER BY just mentions a "list of order comparators" without saying what exactly that is.

In the "Definition: Evaluation of OrderBy" in Section 18.5.2 Evaluation Semantics it is simply called "condition" without any further clarification.

The definition of the corresponding algebra operator also just says "condition" (and the fact that this definition simply states that "the [result] sequence satisfies the ordering condition" is also not particularly helpful).

I have created issue #229 to point out the lack of a definition of "ordering condition"

Tpt · 2025-05-25T19:45:35Z

spec/index.html

+            is an algebraic query expression if
+            |A| is an algebraic query expression.</li>
+          <li id="defn_absGroup">
+            <a href="#defn_absGroup" class="absOp">Group</a>(|exprlist|, |A|)


I am a bit puzzled by having Group here because it can't be composed with any other algebraic query operator (Join, ...) except Aggregation.

Idea: merge AggregateJoin, Aggregation and Group definitions in a single bullet point to make clear they are part of the same construction and not stand-alone and that constructions like Distinct(Group(...)) are not possible.

I was thinking the exact same thing and was planning to get back to this very idea in a follow-up PR. In the given PR, as a first step, I only wanted to make explicit what I see as the algebraic syntax used in the spec so far.

When looking at the corresponding definitions in Section 18.5.2 Evaluation Semantics, it seemed clear to me that "Group(exprlist, P)" and "Aggregation(exprlist, func, scalarvals, Grp)" have been considered as independent operators of the algebraic syntax: they have their own cases/definitions for the eval function (which is formally incorrect because, for these two cases, the result is of a completely different kind than for any other case of the eval function).

Yes! I always got puzzled by these separated definitions for Group and Aggregation. I agree with @afs, it might be nice to have a list of future step and link it here using respec issue feature to make clear to the reader this is a work-in-progress.

I have created issue #230 to point out that 'Group' and 'Aggregation' should not be presented as independent operators of the algebraic syntax, and I will push a commit with inline Notes that point to these GitHub issues using the respec issue feature.

Tpt · 2025-05-25T19:48:27Z

spec/index.html

+            is an algebraic query expression.</li>
+          <li>
+            A sequence of <a href="#defn_sparqlSolutionMapping">solution mappings</a>
+            is an algebraic query expression.</li>


None of the other definitions in this list makes use of the multiset vs sequence distinction. It's weird to have it only here.

The different cases of this overall definition capture any possible type of sub-expression of the algebraic syntax that I have found in the translation algorithm of Section 18.2 Translation to the SPARQL Algebra.

A multiset of solution mappings becomes a sub-expression if the WHERE clause of the given query contains a VALUES clause: see the point "If the form is InlineData" in Section 18.2.2.6 Translate Graph Patterns.

A sequence of solution mappings becomes a sub-expression if the given query contains a trailing VALUES clause: see Section 18.2.4.3 VALUES.

I didn't want to overload the PR by also already including potential changes to the syntax.

Makes perfect sense! Thank you for the detailed answer. I would add this to the list of possible follow ups.

I have added issue #231 to point out that we may want to change this part of the algebraic syntax.

afs · 2025-05-25T21:15:18Z

This PR is the next step...

See the previous PR for a description of the issue: #216

If there are more steps to come, maybe have one issue for this work programme with tasks.

hartig · 2025-05-26T08:55:20Z

If there are more steps to come, maybe have one issue for this work programme with tasks.

Done: #228

…lgebraic syntax

Tpt

Thank you for opening the issues!

hartig · 2025-06-10T06:40:47Z

@afs same question for this PR; are you still planning to review it? Also, in case you haven't seen it, the issue with the work program that suggested me to create, is #228

@rubensworks What do you think of this PR?

w3cbot · 2025-06-19T17:19:22Z

This was discussed during the #rdf-star meeting on 19 June 2025.

View the transcript

w3c/sparql-query#221 and w3c/sparql-query#227

<gb> Pull Request 221 Consider bnode graph names in evaluation of Graph (by hartig)

<gb> Pull Request 227 Adds explicit definition of the algebraic syntax (by hartig)

olaf: I have two PRs on sparql-query that have been opened for a long time
… AndyS, do you still want to look at them?

AndyS: does not look controversial

spec/index.html

afs · 2025-06-23T17:09:03Z

The Echidna step is failing because of HTML validation failure.

 "file:/home/runner/work/sparql-query/sparql-query.w3c/spec/index.html":8343.1-8343.51: error: Element “div” not allowed as child of element “ul” in this context. (Suppressing further errors from this subtree.)

Line numbers are after file inclusion so don't align with the HTML 😞.

The HTML error seems to be these <div class="issue" data-number="..."> in the <ul> list is "18.2 Algebraic Syntax".

hartig · 2025-06-25T11:59:28Z

@afs The Echidna step is failing because of HTML validation failure.

Fixed now.

spec/index.html

afs · 2025-06-25T15:01:54Z

spec/index.html

+        <p id="defn_AlgebraicQueryExpression">An
+          <a href="#defn_AlgebraicQueryExpression">algebraic query expression</a>
+          is defined recursively as follows:</p>
+        <ul>


Presentation comment:, not necessary for this PR:

This might be better as a table. All the <var>A</var> is an algebraic query expressions etc can be written once then the presentation is more compact.

…ary representation.

adds an explicit definition of the algebraic syntax, and uses it in t…

f3e6a17

…he translation of PP expressions/patterns

hartig requested review from Tpt, afs, kasei and rubensworks May 25, 2025 19:10

Tpt reviewed May 25, 2025

View reviewed changes

Tpt approved these changes May 26, 2025

View reviewed changes

hartig mentioned this pull request May 26, 2025

Operators of the algebraic syntax and algebra operators are undistinguishable #228

Open

6 tasks

Notes with pointers to GitHub issues created during the work on the a…

7a5659b

…lgebraic syntax

kasei approved these changes May 26, 2025

View reviewed changes

Tpt approved these changes May 26, 2025

View reviewed changes

w3cbot mentioned this pull request Jun 19, 2025

Consider bnode graph names in evaluation of Graph #221

Merged

afs reviewed Jun 23, 2025

View reviewed changes

spec/index.html Outdated Show resolved Hide resolved

fixes HTML error (div cannot be between li)

f5765fe

afs reviewed Jun 25, 2025

View reviewed changes

afs approved these changes Jun 27, 2025

View reviewed changes

Statement about the translation needs to say that AST is an intermedi…

c945fed

…ary representation.

hartig merged commit 4345c7d into main Jun 27, 2025
2 checks passed

hartig deleted the AlgebraicSyntax branch June 27, 2025 14:13

hartig mentioned this pull request Jun 27, 2025

Adds links to algebraic syntax symbols within the evaluation semantics section #243

Merged

hartig mentioned this pull request Jul 6, 2025

Adds links to algebraic syntax symbols within the translation section #245

Merged

hartig mentioned this pull request Aug 21, 2025

Adds links to algebraic syntax symbols within the Property Paths eval section #268

Merged

Adds explicit definition of the algebraic syntax #227

Adds explicit definition of the algebraic syntax #227

Uh oh!

Conversation

hartig commented May 25, 2025 • edited by pr-preview bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Tpt left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Tpt May 25, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

afs commented May 25, 2025

Uh oh!

hartig commented May 26, 2025

Uh oh!

Tpt left a comment

Choose a reason for hiding this comment

Uh oh!

hartig commented Jun 10, 2025

Uh oh!

w3cbot commented Jun 19, 2025

w3c/sparql-query#221 and w3c/sparql-query#227

Uh oh!

Uh oh!

afs commented Jun 23, 2025

Uh oh!

hartig commented Jun 25, 2025

Uh oh!

Uh oh!

afs Jun 25, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

hartig commented May 25, 2025 •

edited by pr-preview bot

Loading

Tpt May 25, 2025 •

edited

Loading

afs Jun 25, 2025 •

edited

Loading