custom-metrics icon indicating copy to clipboard operation
custom-metrics copied to clipboard

Deprecate the `almanac` custom metric

Open rviscomi opened this issue 2 years ago • 1 comments

The almanac custom metric was created as a catch-all for any new data needed for the 2019 Web Almanac. In subsequent editions, we organized new metrics in "chapter-specific" files. Some almanac metrics are obfuscated by chapter/metric identifiers (my mistake) and some are made redundant by newer custom metrics.

The Web Almanac is on hold this year and its future is unclear. We should decouple the metrics from the project by relocating all of the custom metrics into the most closely related "chapter-specific" files. There should be no loss of data in this migration.

For example, almanac.meta-nodes would be a better fit in the markup custom metric. 10.5 is the name for a structured data custom metric, so it would be a better fit in structured-data. If that custom metric already includes the same functionality, we can drop it.

To help users migrate older queries to newer datasets, we should document where all of the almanac custom metrics went.

rviscomi avatar Jul 21 '23 14:07 rviscomi

Currently, the almanac metric has a lot of useful data that does not seem to be existing in other places. In particular the general node info (number of nodes (e.g., scripts, images) + all attributes on the nodes) was very useful to me (https://github.com/HTTPArchive/custom-metrics/blob/main/dist/almanac.js#L150)

Would be great if this high-level information would be available somewhere and be documented. On har.fyi it currently only says "Metrics defined in the early versions of Web Almanac crawls.".

JannisBush avatar Nov 19 '24 15:11 JannisBush