fix bug in Rakefile after removing common/

[pchampin]

No description provided.

[w3cbot]

This was discussed during the #json-ld meeting on 24 June 2026.

View the transcript

w3c/json-ld-api#693

pchampin: this PR is doing two things: fixing an issue created by the mutualization of common files
… but also commenting out the checking all the tests in the spec;
… the reason for commenting it out is that the tools we have currently do not support JSON-LD 1.2, which the examples are
… we can not rely on our tools being constantly up-to-date with the spec

anatoly-scherbakov: IIUC, the problem is the Ruby implementation lacking a maintainer;
… what would be the process to chose another implementation

TallTed: if we comment out, we should open an issue to keep track of this

niklasl: anatoly-scherbakov, good question, I'm conflicted.
… I would recommend my own tool, CONS: it is a one-man-show, no other contributors,
… PROS: no dependency (no dependency rot, not induced security issue)

pchampin: anatoly-scherbakov, the problem is not so much chosing another impl, but relying on it being constantly up-to-date
… for the dependency issue, dependabot is doing a good job

anatoly-scherbakov: the absence of implementations should not impede the development of the spec;
… we could run the tests in a way that provides a dashboard, but does not block PRs to be merged

bigbluehat: yes, that would be useful; anatoly-scherbakov could you work on that?

anatoly-scherbakov: yes, I could do that with PyLD, with which I'm more familiar

bigbluehat: I don't want to frame this as a change of tooling, though I'm not opposed to it
… the main goal is to solve the race condition we have

<anatoly-scherbakov> +1

<niklasl> +1

bigbluehat: let's not rush to solution; let's merge the PR to unlock other PRs, and discuss this further

<pchampin> +1

<anatoly-scherbakov> +1

bigbluehat: a first action would be to tweak the GH action to be non-blocking, regardless of the implementation
… then decide on switching to a different implementation

<niklasl> Apart from already having 671 in, TRLD is apparently bug-for-bug compatible with the spec even. ;D

---