Improving Project discovery

[GBH]

Right now project search is basically: choose language and scroll until you can't through random (literally order("RANDOM()")) list of project names. That list doesn't tell me anything about the project. There may or may not be a contrubulator score that tells me literally nothing. It's always in 20-30 range? Is that bad or good? Clicking on it sometimes takes me to some page that tells me it was last updated 2 years ago. Hardly useful or confidence inspiring.

Here are few things that would be great to see:

• Project's number of stars/forks/contributors
• Number of open issues and PRs (maybe against closed ones to see the ratio)
• Timestamp of the last commit

As far as listing goes:

• Filtering actions should be reflected in address bar.
• Should display how many projects are being displayed.
• Really dislike infinite scroll as you'll never know where you are in the list
• Allow sorting by recent activity, number of issues, number of stars, etc

This means that process of adding projects should be reworked a bit.

• Only Github URL needed.
• API Call to pull in everything about the project
• Language stats can be pulled in via API call, that would resolve Allow user to specify more than one language #1002
• Anybody can refresh project stats at any time (and maybe nightly cron)
• Project doesn't belong to whoever "suggested" it first.

I wouldn't mind tackling some of this. What do you think?

[niraj8]

Came here to see if the project is in active development, as the site hasn't changed much since last year. I also had some of the concerns regarding freshness of the repo suggestions.

Good suggestions on adding more info about the repo to the display, shouldn't take a lot of effort.

I'd like to take up some tasks as well.

[BenJam]

@GBH thanks for proposing this!

It seems like there's two threads to this: A wider appreciation of the quality of a project, and the more tactical ability to make some of those judgements yourself within a minimal set of data?

On the first point: We (@andrew and I) been thinking about many of the same things from a Libraries.io perspective (that work is over here https://github.com/librariesio/metrics). OSS Cert have been doing something similar too https://ossert.evilmartians.io/split#maturity

Using Libraries.io as an agnostic source of data on projects would certainly give us plenty of rope for beginning this. From there we could decide what metrics would be useful to turn into measures, whether that's a single figure or something more granular (perhaps a code, community, documentation, support split?).

On the tactical side: Maybe the first step along that path might be to show the data you think is useful right in those results?

[GBH]

I honestly don't think there's a need of sophisticated project metrics. The only thing that matters is the fact that somebody uses it and that PR has a chance to be merged. Basic stats like that can be easily extracted from Github.