search

digipres / digipres.github.io

Auto-generated static web site digipres.org
https://www.digipres.org/
26 stars 19 forks source link

Extension counts in format radar charts and registry comparisons disagree #15

Closed nkrabben closed 2 years ago

nkrabben commented 7 years ago

For example, radar chart for LC FDD says 235... http://www.digipres.org/formats/sources/fdd/ ...and the overlap chart sums up to >250 http://www.digipres.org/formats/overlaps/

It's not clear which count is accurate.

anjackson commented 7 years ago

IIRC, I thought at the time that showing the number of format entries that had one-or-more extensions made more sense (i.e. you could get a sense of whether the registry tried to get all properties for all formats, because ideally every format might have an extension & MIME type & magic (which would be a square plot).

However, looking at it now, it just seems rather confusing.

anjackson commented 2 years ago

To clarify, the radar chart shows the number of records that have a file extension, and the comparisons show the number of unique file extensions. Many records have the same file extension (e.g. file versions), so this is why they appear to disagree.

I think it makes more sense to use the unique extension and MIME type counts everywhere.

anjackson commented 2 years ago

Okay, I think that's implemented properly now.