Repository navigation
Conversation
Embed LDForge's editor and result renderers while retaining Fuseki's dataset navigation, endpoint selection, example queries and prefix badges. Use the Fuseki logo color for the editor theme and preserve query sharing through the editor toolbar. Use published LDForge packages, update documentation and query tests, and cover editor resizing, prefix toggling and query execution. Validation: production build, 52 unit tests and 18 Cypress end-to-end tests passed.
Pin core 1.0.1, editor 1.2.1, geo/renderers 1.1.1 and transitive prefixes 1.0.1. Verify packaged MIT license files, record dependency inventory and add reproducible tests against real Fuseki 6.0.0. Remove the additional SPARQL editor README section. Validation: production build, 52 unit tests, 18 mock-backed Cypress tests and 7 real-server Cypress tests passed.
|
The goals are interesting. What isn't mentioned is that LDForge is written by the submitter and it is apparently 3 days old. (2026-11-13) The license is "MIT with attribution" (it changed 2 hours ago from EUPL). Jena releases every 3-4 months (or longer - releases happen if people-time allows). I'd like to see evidence of usage.
Would it be possible to work with Zazuko to update to CodeMirror 6? |
|
I’d like to provide some additional background on LDForge and my motivation for submitting this PR. First, to be fully transparent: I am both the creator of LDForge and the author of this pull request. I have a long history with Yasgui. I’ve worked with it extensively over the years, and I also worked at Triply, the company behind its original development. I have a great deal of respect for what Yasgui has contributed to the SPARQL community. It has been an essential tool for many years. However, I’ve also been considering building an alternative for quite some time. Yasgui is showing its age, not just because of its reliance on CodeMirror 5, but also because of its architecture, limited extensibility, and plugin system. The original plugin documentation is no longer properly available, making it increasingly difficult for developers to build upon it. LDForge is not intended to be just Yasgui with CodeMirror 6. My goal is to rethink the entire approach: a modular, extensible, developer-friendly SPARQL IDE built around modern JavaScript/TypeScript practices, with well-defined APIs, independently reusable components, and a much better developer experience. Although LDForge itself is a relatively new open-source project, it originated as a spin-off from LDMax, my Linked Data platform. Its underlying concepts and components have already been developed and used extensively within that platform. So while the standalone project is new, it is not something I started from scratch last week. I also want to explain why I’m proposing this integration for Fuseki specifically. Fuseki is an important part of the SPARQL ecosystem, and I believe integrating LDForge here could help accelerate its adoption, attract contributors, and establish a sustainable modern alternative to Yasgui. Given the age of the existing tooling, I believe this would ultimately benefit the wider community, not just LDForge or Fuseki. Finally, regarding licensing: I specifically switched LDForge to the MIT license to make this contribution possible and remove potential licensing concerns. At this early stage of the project, I don’t have a strong preference for a more restrictive license. Making LDForge accessible, useful, and easy to adopt is much more important to me. I fully understand that replacing an established component in Apache Jena is not a trivial decision, especially with a relatively new project. I’m happy to discuss concerns about maturity, maintenance, compatibility, or long-term sustainability, and to work with the maintainers to address them. My intention is to contribute something useful to the SPARQL community and, hopefully, help move its developer tooling forward. |
|
The big issue I see here is governance. With yasgui moved under rdfjs I think there's a more clear path for the community to keep maintaining it (there's an open PR for code mirror 6), and there are more voices, eyes, and time on that project than LDForge. I think it's an interesting idea, but switching off of yasgui for a newly open sourced enterprise sponsored solution is a hard sell. |
|
I understand your concerns, especially regarding governance and long-term sustainability. I have absolutely no objection to moving the entire LDForge stack under RDFJS governance. In fact, I would welcome it. My goal is not to maintain control over yet another SPARQL editor, but to provide the RDF community with a modern, well-maintained and developer-friendly tool for writing and executing queries. That said, I have some concerns about the current state of the [RDFJS Yasgui repository](https://github.com/rdfjs/Yasgui). While moving Yasgui under RDFJS governance is a positive step, it doesn't automatically resolve the underlying issues. Documentation links are broken, and there seems to be some confusion about the roles of Zazuko and Triply in the project's development and maintenance. I also want to emphasize that my motivation goes beyond upgrading CodeMirror. I believe the entire architecture and developer experience deserve a fresh approach, particularly when it comes to extensibility, plugins, documentation and integration into other applications. I have the advantage of knowing Yasgui and its history quite well. I've worked with it for many years and also worked at Triply alongside its original developers. I know how valuable Yasgui has been to the community, but I'm also familiar with its limitations. LDForge may be newly open-sourced, but it isn't an attempt to introduce a commercially controlled alternative. It grew out of my work on LDMax, and I would be perfectly happy to see it evolve into a community-governed project, independent of LDMax. Ultimately, I think we're aiming for the same thing: a sustainable, modern SPARQL query interface that the entire RDF community can rely on. If bringing LDForge under RDFJS governance helps achieve that, I'm more than willing to explore that path. |
|
There's a bigger concern here that's been raised on several similar large proposed contributions lately (e.g. #4363, #4364) about how something like this is going to be maintained longer term, especially when they come from contributors with little/no prior history of engagement with the project. Also as others have pointed out, as with other recent proposals, this appears sufficiently early stage/lacking community (from our perspective) to consider for inclusion in Jena at this time Often it is better for new things like this to build their own community first before proposing to integrate into Jena, if that's even the right path for them long term. Jena is entirely volunteer driven, and none of the PMC have any meaningful paid time on the project, as Apache projects go it's about as far from the corporate driven headliners with full time paid developers (Hadoop, Spark, Kafka etc) as you can get. We all have very limited time to work on Jena, which includes reviewing outside contributions. When contributors are increasingly showing up with no prior history of engagement with the project and suggesting large sweeping changes to the project then the existing committers are naturally going to be skeptical and disinclined to review, yet along accept the contribution. |
Yasgui has served the SPARQL community well, but its original upstream repository is now archived. Fuseki currently uses the maintained RDFJS/Zazuko packages, which continue to provide fixes, but the query editor still relies on CodeMirror 5. This leaves a central part of the Fuseki user interface dependent on an older editor architecture.
I believe it is time to move towards a modern SPARQL IDE rather than continue extending that integration. This PR proposes LDForge as that replacement and provides a working integration for review.
LDForge uses CodeMirror 6 and separates query execution, editing and result rendering into independently usable packages. This gives Fuseki a foundation for developing its query interface further, including contextual completion, diagnostics, query tabs and additional result views.
The integration retains Fuseki’s existing dataset navigation, explanatory text, example queries, editable SPARQL endpoint and prefix buttons. Prefix buttons reflect the active query and support adding and removing declarations without moving the cursor. The editor also supports resizing, cancellation, query sharing and CSV downloads.
The editor follows the blue from the Fuseki logo through its theme API. Its JavaScript and CSS are bundled with the application from pinned npm dependencies; no local development checkout is required. This uses editor 1.2.1, core 1.0.1, geo/renderers 1.1.1 and transitive prefixes 1.0.1.
Table, boolean, JSON and geographic result views are included. This is not a complete reproduction of every Yasr feature: the previous response-format selectors are not included, and geographic rendering uses LDForge’s binding conventions. These differences should be considered during review.
Validation completed:
The real-server test fixtures and reproduction instructions are included in
jena-fuseki2/jena-fuseki-ui/tests/integration/. External map/CDN loading and deployment-specific authentication configurations are outside that suite.