Also important is the extent to which the maintainers are willing and able to change the library to respond to changes in users' needs and expectations. Parsec, for example, seems to have been largely supplanted by Megaparsec because it failed to evolve. And some libraries need to track developments in implementation techniques, either replacing old techniques or adding new ones alongside. On Sep 12, 2016 2:53 PM, "Joachim Durchholz" <jo@durchholz.org> wrote: The best signal is a compact description of what the library does, and, more importantly, what it does not do. Stability isn't a particularly interesting metric anyway. You want metrics like use case coverage, interface complexity, performance predictability, and you also want the update intensity over time (the shape of the curve correlates with kinds of team activity - you need to validate anything you infer from the curve, but it tells you what things to validate). Am 12.09.2016 um 11:49 schrieb Theodore Lief Gannon:

...and, "it will probably be replaced by something meaningful in the future." But it hasn't, nor has it been removed, and it's a better signal than no signal at all.

On Sun, Sep 11, 2016 at 4:01 AM, Tom Ellis < tom-lists-haskell-cafe-2013@jaguarpaw.co.uk> wrote:

On Sun, Sep 11, 2016 at 03:48:38AM -0700, Theodore Lief Gannon wrote:

...

...

In fact, there's a "Stability" field where you can indicate that it's officially experimental.

Six years ago Simon Marlow pronounced that the "Stability" field was "mostly defunct now, and shouldn't be used"

https://stackoverflow.com/questions/3841218/conventions- for-stability-field-of-cabal-packages/3847493#3847493

Tom

_______________________________________________ Haskell-Cafe mailing list To (un)subscribe, modify options or view archives go to: http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe Only members subscribed via the mailman list are allowed to post.