Conversation
…eter The N-Triples serializer writes statements in the order the store iterates them, which follows set iteration and so varies between processes. The same graph therefore serializes to different bytes in different runs, which shows up as diff noise in version-controlled RDF. PR RDFLib#3008 solved this for longturtle with an opt-in `canon` parameter that canonicalizes the graph and sorts, fixing RDFLib#1890. This gives NTSerializer the same parameter, with the same default of False, so ordinary serialization is untouched and still streams statement by statement. `canon=True` canonicalizes with to_canonical_graph and writes the rendered rows in sorted order. Sorting the rendered rows rather than the triples keeps the order total: two distinct terms can share a str(), which would leave their relative order down to the store iteration order that this is meant to remove, whereas identical rows mean identical statements. Measured over six PYTHONHASHSEED values on a graph with a hub, a shared list tail and a deep blank-node chain: 5 distinct documents before, 1 after. NT11Serializer inherits it by subclassing.
This was referenced Sep 14, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Mirror of RDFLib#3546 — tracking only. Do not merge.
Upstream PR: RDFLib#3546
Why this PR exists
Every fix this organisation depends on is written as a self-contained, single-commit
upstream contribution rather than a local patch, so that a fork is only needed for as
long as its contributions are unmerged.
That convention has one weakness: an upstream PR is easy to lose track of, because it
lives in someone else's repository and closes on someone else's schedule. This mirrored PR
is the counterweight — an open PR in our own fork for every contribution still pending
upstream, so the set of things we are carrying is visible from our own repository list.
It is closed when the upstream PR is merged or rejected, not before.
Contents
A single commit, identical in content to the upstream PR head:
feat/deterministic-ntriples->mainWhere it is used
Not yet consumed: the pipeline pins released
rdflib==7.6.0. Relevant todiffable-rdf, which re-serializes canonicalized graphs through rdflib.