Tracking JavaScript library versions in HTTP archive

Analysis
21

Sweet. Couple of questions…

  1. Where should these scripts live? A subdirectory under main WPT repo, or a separate repo under WPO-Foundation org?
  2. What’s the required plumbing to get these scripts to run at the end of each page load?
  3. Is there any extra plumbing to get the emitted values into the current HA dumps? I’m using the CSV-formatted files for BQ import.

@sam would you be willing to put together a first run at a framework-version script? :smile:

1 Like
22

The bigger issue is changing the DB schema and making sure the scripts don’t extend the crawl time significantly. Those changes require intimate knowledge of the code.

23

If you can sort out the plumbing I would be super happy to configure the detection script and review detection PRs. I am not concerned about crawl times cause the detections are very very cheap.

We would have to be super diligent not to merge in anything fancy.

24

In the news today: http://libscore.com/. Background: https://medium.com/@Shapiro/introducing-libscore-com-be93165fa497 - I don’t see version numbers in their dataset, but it seems like they’re 99% of the way there, and it wouldn’t be too hard to add?

That’s not to say HA couldn’t or shouldn’t implement what we discussed above – in fact, personally I think we still should – but just FYI.

2 Likes
26

For anyone interested, I’m tracking this feature on GitHub at https://github.com/HTTPArchive/httparchive/issues/77