Node ID and OR filter bug by miratepuffin · Pull Request #2754 · Pometry/Raphtory

miratepuffin · 2026-09-01T14:01:21Z

What changes were proposed in this pull request?

Fixes three filter-correctness bugs in Raphtory, all found while property-testing filter equivalence, and makes a filtered-read denial legible.

Edge leak in or filters (filter.rs) — an or-filtered node list was wrongly treated as trusted, so the edge iterator's fast path kept edges to excluded nodes.
.filter() on a node collection dropped members (nodes.rs) — nodes.filter(name==…) narrowed membership; it should keep every member (that's what nodes[…] is for). Membership is now read from the base graph, consistently across fields.
A time view on an edge collection was ignored (edges.rs) — edges[before(t)] (and after/window/at/latest) returned every edge instead of narrowing. It combined the filter against the unwindowed base graph, dropping the view; it now chains onto the current view like node collections do.

Plus: a filtered-read grant whose resolved filter can't be applied now denies with a clear, logged message instead of an opaque "internal error" (data.rs).

Plus: config-schema introspection — GraphServer.config_schema() in Python and --print-config-schema on the CLI return the full config schema as JSON: every field the server and every registered extension accept, including ones unset by default (print_config only shows set values). Backed by a new config_schema hook on server extensions (defaults to to_json).

Why are the changes needed?

All three produced silently wrong views; (3) failed open (an edge collection filtered by time processed every edge). None were caught because existing tests only assert on .id and only hit the paths that happened to work.

Does this PR introduce any user-facing change? If yes is this documented?

Behavioural fixes, no API change:

nodes.filter(name=='a')       -> ['a','b']   (was ['a'])
nodes[name=='a']              -> ['a']
edges[Graph.before(t)]        -> narrowed     (was every edge)

Behaviour now matches the documented .filter vs [...] semantics; no doc changes needed.

New API surface: GraphServer.config_schema() (Python, documented in the stub) and the --print-config-schema server flag (documented in --help).

How was this patch tested?

New unit and Python collection tests (or/and-of-or edge-endpoint invariants; node-collection membership; edge-collection filtering across every filter type and time view; access-filter denial logging), plus the full filter regression suite — all pass.

Are there any further changes required?

Yes, tracked separately: time views don't compose reliably on edge collections — a pure time view is fixed here, but combining one with another filter (and/or/not) can still drop it and fail open. Same class as (3), in the composite filter path. Worth a parametrised suite over {collection path} × {filter type} × {and/or/not}, which currently has large gaps.

…lence.

ljeub-pometry

the or filter is now too pessimistic about its const-ness (which is correct but inefficient)

ljeub-pometry · 2026-09-01T14:05:28Z

    return check


+def _combo_offenders(graph, op, join):


better name...

ljeub-pometry · 2026-09-01T14:06:28Z

+        # The shape behind the RBAC leak: a node-set restriction ANDed with a wide `or` that
+        # includes a node_type branch (whose domain is every node).


clean up internal comments

github-actions

⚠️ Performance Alert ⚠️

Possible performance regression was detected for benchmark 'Rust Benchmark'.
Benchmark result of this commit is worse than the previous benchmark result exceeding threshold 2.

Benchmark suite	Current: `9881498`	Previous: `9823ef7`	Ratio
`lotr_graph/num_edges`	`5` ns/iter (`± 0`)	`0` ns/iter (`± 0`)	`+∞`
`lotr_graph/num_nodes`	`4` ns/iter (`± 0`)	`1` ns/iter (`± 0`)	`4`
`lotr_graph/graph_latest`	`2` ns/iter (`± 0`)	`0` ns/iter (`± 0`)	`+∞`
`lotr_graph_materialise/materialize`	`6571473` ns/iter (`± 33378`)	`1564816` ns/iter (`± 35303`)	`4.20`
`lotr_graph_window_100_materialise/materialize`	`6591879` ns/iter (`± 60406`)	`1669150` ns/iter (`± 10700`)	`3.95`
`lotr_graph_window_10_materialise/materialize`	`2711230` ns/iter (`± 10084`)	`971980` ns/iter (`± 4278`)	`2.79`
`lotr_graph_subgraph_10pc/num_nodes`	`15` ns/iter (`± 0`)	`4` ns/iter (`± 0`)	`3.75`
`lotr_graph_subgraph_10pc_materialise/materialize`	`1801590` ns/iter (`± 27769`)	`334634` ns/iter (`± 1287`)	`5.38`
`lotr_graph_subgraph_10pc_windowed_materialise/materialize`	`1080459` ns/iter (`± 9677`)	`230399` ns/iter (`± 2617`)	`4.69`
`lotr_graph_window_50_layered/has_node_existing`	`318` ns/iter (`± 23`)	`129` ns/iter (`± 12`)	`2.47`
`lotr_graph_window_50_layered_materialise/materialize`	`24688256` ns/iter (`± 320228`)	`3488825` ns/iter (`± 24948`)	`7.08`
`lotr_graph_persistent_window_50_layered/num_edges_temporal`	`494682` ns/iter (`± 7891`)	`192686` ns/iter (`± 1569`)	`2.57`
`lotr_graph_persistent_window_50_layered_materialise/materialize`	`41046675` ns/iter (`± 186306`)	`5298035` ns/iter (`± 147912`)	`7.75`

This comment was automatically generated by workflow using github-action-benchmark.

miratepuffin · 2026-09-02T10:47:48Z

+
+
+def _edges(collection):
+    return sorted((e.src.name, e.dst.name) for e in collection)


This can be e.id

miratepuffin · 2026-09-02T10:54:07Z

-    narrow collection membership — unlike the property filter above. Pins the
-    new behavior (and its parity with local) so a regression is visible."""
+def test_filter_by_node_id_keeps_membership_but_getitem_narrows():
+    """`.filter()` keeps every member for a name/id filter too — `[]` is the one that narrows.


.filter only filters to the right, can also add the case where we do g.filer.nodes.id here

miratepuffin · 2026-09-02T10:54:59Z

    }

+    /// The full config schema as a nested dict: every field, including ones unset by default.
+    ///


Need to add this to the PR description

miratepuffin · 2026-09-02T10:55:41Z

-            if print_config {
+            if print_config_schema {
+                let schema = server.config().config_schema_json().map_err(|e| {
+                    std::io::Error::new(std::io::ErrorKind::InvalidData, e.to_string())


tidy this up

miratepuffin · 2026-09-02T10:59:52Z

    }

    pub fn node_list(&self) -> NodeList {
+        // Membership is the candidate set (`self.nodes` / the base graph), filtered per node by


don't need to this comment

…li schema print, drop comment Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Co-authored-by: Claude Fable 5 <noreply@anthropic.com> Co-authored-by: Shivam <4599890+shivamka1@users.noreply.github.com> # Conflicts: # raphtory/src/db/graph/nodes.rs

Node-id equality and set membership resolve their evaluation domain through the storage index instead of visiting every node, carrying the narrowing #2754 added to the removed builder op onto expressions. A constant whose type does not match the graph's id type keeps the full domain rather than guessing, and every other predicate is unchanged. A sum widens at runtime but declared its element type, so a constant beyond that element's range was rejected before the sum ran — no comparison above 255 was expressible on a list of u8. Sum now declares what the evaluator produces; reductions returning an element keep the element type. The generated stubs and the docstrings pointing at them still named builder classes that no longer exist.

Two node-filter bugs, both found while property-testing filter equiva…

1e0f86e

…lence.

ljeub-pometry requested changes Sep 1, 2026

View reviewed changes

github-actions Bot reviewed Sep 1, 2026

View reviewed changes

miratepuffin and others added 6 commits September 1, 2026 17:41

Quick update

c5a3ad4

expose conf on server

50f001f

Modifying message which is correctly logged

9129324

Fix issues with edges

93a3202

fmt

454f042

chore: apply tidy-public auto-fixes

e34805a

miratepuffin commented Sep 2, 2026

View reviewed changes

miratepuffin and others added 4 commits September 2, 2026 12:07

Review comments: e.id in test helper, graph-level filter case, tidy c…

7e36cab

…li schema print, drop comment Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

add tests for edge filters

fed684b

Merge branch 'db_v4' into or-filter-bug

9a7dc73

Merge branch 'db_v4' into or-filter-bug

64a4f3f

shivamka1 approved these changes Sep 3, 2026

View reviewed changes

chore: apply tidy-public auto-fixes

9881498

miratepuffin merged commit 06e85d9 into db_v4 Sep 3, 2026
32 of 33 checks passed

miratepuffin deleted the or-filter-bug branch September 3, 2026 20:08

This was referenced Sep 4, 2026

Type-safe filter expression API #2643

Draft

Aggregate is_deleted over an edge's layers #2767

Open

Check filter composition against set algebra #2768

Closed

Navigation Menu

Uh oh!

Node ID and OR filter bug - #2754

Node ID and OR filter bug#2754
miratepuffin merged 12 commits into
db_v4from
or-filter-bug

miratepuffin commented Sep 1, 2026 •

edited

Loading

Uh oh!

ljeub-pometry left a comment

Uh oh!

ljeub-pometry Sep 1, 2026

Uh oh!

ljeub-pometry Sep 1, 2026

Uh oh!

Uh oh!

Uh oh!

Uh oh!

github-actions Bot left a comment •

edited

Loading

Uh oh!

miratepuffin Sep 2, 2026

Uh oh!

miratepuffin Sep 2, 2026

Uh oh!

miratepuffin Sep 2, 2026

Uh oh!

miratepuffin Sep 2, 2026

Uh oh!

miratepuffin Sep 2, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

		# The shape behind the RBAC leak: a node-set restriction ANDed with a wide `or` that
		# includes a node_type branch (whose domain is every node).



		def _edges(collection):
		return sorted((e.src.name, e.dst.name) for e in collection)

Uh oh!

Conversation

miratepuffin commented Sep 1, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What changes were proposed in this pull request?

Why are the changes needed?

Does this PR introduce any user-facing change? If yes is this documented?

How was this patch tested?

Are there any further changes required?

Uh oh!

ljeub-pometry left a comment

Choose a reason for hiding this comment

Uh oh!

ljeub-pometry Sep 1, 2026

Choose a reason for hiding this comment

Uh oh!

ljeub-pometry Sep 1, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

github-actions Bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

⚠️ Performance Alert ⚠️

Uh oh!

miratepuffin Sep 2, 2026

Choose a reason for hiding this comment

Uh oh!

miratepuffin Sep 2, 2026

Choose a reason for hiding this comment

Uh oh!

miratepuffin Sep 2, 2026

Choose a reason for hiding this comment

Uh oh!

miratepuffin Sep 2, 2026

Choose a reason for hiding this comment

Uh oh!

miratepuffin Sep 2, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

miratepuffin commented Sep 1, 2026 •

edited

Loading

github-actions Bot left a comment •

edited

Loading