Broken references in MPEG-2 TS Byte Stream Format

[github-actions]

This pull request was automatically created by Strudy upon detecting errors in MPEG-2 TS Byte Stream Format.

Please check that these errors were correctly detected, and that they have not already been reported in https://github.com/w3c/mse-byte-stream-format-mp2t.

If everything is OK, you can merge this pull request which will report the issue below to the repo, and update the underlying report file with a link to the said issue.

---

Repo: 'https://github.com/w3c/mse-byte-stream-format-mp2t'
Tracked: N/A
Title: Broken references in MPEG-2 TS Byte Stream Format

While crawling MPEG-2 TS Byte Stream Format, the following links to other specifications were detected as pointing to non-existing anchors:

• https://www.w3.org/TR/media-source-2/#init-segment
• https://www.w3.org/TR/media-source-2/#media-segment
• https://www.w3.org/TR/media-source-2/#coded-frame

_{This issue was detected and reported semi-automatically by Strudy based on data collected in webref.}

[tidoust]

FWIW, this and the other MSE bye stream format issues are the usual TR vs. ED problem. The broken links do not exist in the Editors Drafts, which haven't been republished as /TR Notes yet because a few other editorial updates are needed for that to be possible.

[tidoust]

So... this and the other MSE byte stream format issues were not the usual TR vs. ED problem ;)

What happened is the following:

0. Starting point: The MSE had explicit IDs for some of the terms it exported, e.g., init-segment. These IDs were in Webref (and thus in the xref database). These IDs were also hardcoded in the MSE byte stream format specs.
1. The MSE byte stream format specs were first updated to use usual cross referencing syntax. This triggered the update on 6 December. Note the specs served from GitHub Pages contain the source of the spec, not the generated spec (no spec-prod action used for now).
2. The crawler noticed the update (new last modified date) and refreshed the data in Webref accordingly. But at that point in time, the MSE byte stream format specs on GitHub Pages still referenced the explicit IDs, since that's what the MSE and the xref database contained. So #init-segment, etc.
3. The MSE spec was updated to drop the explicit IDs. New IDs were assigned to the terms as a result.
4. The crawler refreshed the MSE spec extracts and the xref database to use the new IDs.
5. The MSE byte stream format specs automatically pick up the new IDs but since the version served from GitHub Pages is the source, and since the source did not change, this does not trigger any change to the last modified date.
6. Reffy happily skips the MSE byte stream format specs because the last modified date hasn't changed. So the links extracts for these specs in Webref remain incorrect.

In short, the problem is that the last modified date cannot be trusted for cross-references links when we crawl the source of a ReSpec spec because the xref database might have changed under the hoods.

[dontcallmedom]

ha, thanks for digging into this, interesting bug!