Metadata error isn't helpful

[jyasskin]

https://lists.w3.org/Archives/Public/public-tr-notifications/2017Sep/0104.html says:

On Mon Sep 25 20:45:02 UTC 2017, the request to publish https://raw.githubusercontent.com/w3c/permissions/master/.echidna failed. See details below.
...
  "results": {
    "status": "error",
    "jobs": {
      "retrieve-resources": {
        "status": "ok",
        "errors": []
      },
      "metadata": {
        "status": "error",
        "errors": [
          "Error: Not Found"
        ]
      },
...
      {
        "time": "2017-09-25T20:45:02.554Z",
        "fact": "An error occurred while running the metadata extractor."
      }

This doesn't help me figure out how to fix the metadata.

[tripu]

@jyasskin, the URL that is passed to Echidna should be either the spec to publish, or a manifest. The way Echidna discerns which one it is is by checking whether the first char is < (spec) or not (manifest).

I think in this case you should pass the actual URL of the spec (https://api.csswg.org/bikeshed/?url=https://raw.githubusercontent.com/w3c/permissions/master/index.bs&md-status=WD&md-prepare-for-tr=true), not the URL of a file containing the URL of the spec (https://raw.githubusercontent.com/w3c/permissions/master/.echidna).

[jyasskin]

Please say that in the error message.

[tripu]

“Please say that in the error message.”

@jyasskin, I'm not sure that checking for this specific type of mistake makes sense in Echidna. The URL posted to Echidna must be either the spec itself, or a manifests (ie, a list of relative links).

The URL submitted could be anything else by mistake (a Git repository, an image). I'm not sure Echidna should check for all possible errors, and report specifically about those.

An analogy to illustrate my point: when I submit to Bikeshed a URL that is not a Bikeshed source file (as it should be), I don't get an error saying so. I just get a generic error (in this case, not even so; just a blank page).

[jyasskin]

The file here was detected as a manifest (since https://raw.githubusercontent.com/w3c/permissions/master/.echidna doesn't start with a <) but then didn't parse as a manifest (since the URL was absolute). That failure to parse seems like the right thing to report.

https://raw.githubusercontent.com/w3c/permissions/master/.echidna is not a valid manifest (https://github.com/w3c/echidna/wiki/Preparing-your-document#manifest-file)

might be a good error. Talking about a "metadata extractor" makes it seem like I should be looking at https://github.com/w3c/echidna/wiki/Metadata-extracted-from-the-document, which implies that you've already found a spec document using my manifest, which wasn't true.

[tripu]

Okay, @jyasskin.
I'll try to work on better errors for this specific issue (a manifest containing an absolute URI).