"Sort by" feature

[MangoIV]

It would be nice to have a "sort by" feature, i.e. something like

• best match (the standard)
• downloads
• outdatedness

and a modifier for ascending/ descending

[jappeace]

specifcally for anwsering the question of "is this package being maintained", it'd be really nice if you know that before having to go trough compiling it and finding issues like bounds.

[TixieSalander]

For now release dates are display in the versions list in the package "single page" but indeed having the last version release date present directly in packages lists and being able to sort lists on that criteria is definitely useful and will be probably be added.

[TixieSalander]

@MangoIV I agree! @tchoutri recently mentioned this feature to me as something planned in the future so it very should be added.

At the exception of downloads since we currently can't get accurate data on that:

1. Hackage have indeed a download count but the number isn't accurate at all. It doesn't include CDN who put package in cache (and other ways of getting a package, If I remember correctly what @tchoutri told me)
2. Other repositories aggregated by Flora doesn't necessary provide download data

But date related sorting on packages lists is a good thing indeed! 👍

[tchoutri]

@MangoIV Still that wouldn't work because I imagine that highly-requested packages stay "hot" in the CDN, thus getting few hits on Hackage. But then how do you distinguish packages that are scarcely requested and those so requested that the CDN keeps them all the time? ;-)

Unless hackage.haskell.org is patched to get fed CDN data, download numbers will be unreliable.
(moreover, we also index Cardano and Horizon, which have no data at all, since they are static sites)

[tchoutri]

That being said a stand-alone installation of Flora could certainly do this.

[MangoIV]

Still that wouldn't work because I imagine that highly-requested packages stay "hot" in the CDN, thus getting few hits on Hackage. But then how do you distinguish packages that are scarcely requested and those so requested that the CDN keeps them all the time?

so do you think that there’s a package with some amount of real downloads and an amount of hackage downloads, such that a package with more hackage downloads has less or equal than real downloads? Because if not that wouldn’t be a problem.

[tchoutri]

I don't understand how this connects to my above statement

[MangoIV]

In the way that you would still capture the proportions correctly like proposed in the original answer.