After a long gap, a new META 100 list for the Top 100 site

Alias on 2009-08-19T17:46:34

It's been a while since I added a new list to the Top 100 website.

This is partly because I didn't want to distract from the focus on cleaning up the FAIL 100 list, and partly because it's very hard to come up ideas for metrics (and lists) that won't generate bad feedback loops.

But one problem I have noticed while working on my graph development is that there are various gaping holes in the data. Some nodes appear as having no dependencies when really they just have a broken or bad META.yml.

And fixing this distortion is going to be important for the health of the graphing, analysis and weighting of every other list.

So I've added a new META 100 list, to identify missing or broken META.yml files in the CPAN (in volatility order of course, so we can prioritise the order in which these modules should be fixed).

Of course, it's mostly likely not going to be without controversy, because although the META.yml specification SAYS that it's written in YAML it's not really true. From a Parse::CPAN::Meta perspective, it's written in the "YAML Tiny" subset language. So people that have taken the word "YAML" in the spec as license to drop JSON into their YAML files (using on the whole "JSON is a subset of YAML" reasoning) are going to be rather disappointed to find themselves listed as "broken".

But then this fight is mostly over at this point. If you want JSON then we're going to give you JSON in the form of META.json after 5.10.1 is released. And then the META.yml spec will (most likely) be updated to refer specifically to the subset language, rather than just "YAML".

(There are also some older conventions in META.yml that aren't strictly correct in YAML anymore anyway that Parse::CPAN::Meta supports as a special case, but that's another story).

In other top 100 news, I've now very close to integrating the RT ticket data with CPANDB, and I've got what I hope is a fairly solid concept for a "Buggy" distribution metric that balances the different factors properly... although the name could be better. The Tickety 100 perhaps?