Thanks Teo, that's exactly what I was looking for
I'll keep in mind this number changes over time.
@Andrea I think the intention is that the patches eventually get
upstreamed to the libraries
It's just a good measure point for me to do quantitative breakage
analysis upon.
Hi Jappie,
There is a GHC grafana with some statistics. Unfortunately the relevant dashboards seem to be broken at present. Hopefully someone on this list could fix it.
There's also https://grafana.gitlab.haskell.org/d/RHebDhq7z/head-hackage-residency-profiling?orgId=2&refresh=30m, which is for residency information, but looking at the filters suggests that the peak amount of packages is around 900.
Cheers,Teo
On Tue, Jul 23, 2024 at 4:29 AM Andrea Bedini <andrea@andreabedini.com> wrote:
On Tue, 23 Jul 2024, at 7:38 AM, jappie klooster wrote:
> For the stability working group I'm trying to understand how many
> packages head.hackage allows you to build.
> Does anyone know the answer to this or have an idea how I can find out?
I don't have the numbers you are looking for but one data-point is that cabal-install itself (which has a reasonable number of dependencies) typically does not compile with a freshly released GHC. This forces to temporarily add head.hackage in CI when a new GHC releases.
IMHO, using head.hackage to test a source-distribution defeats the purposes of testing. We cannot (and should not) assume those building cabal's source distribution(s) are using it, and therefore `cabal install cabal-install` will end up succeeding in our CI and failing for everybody else.
My 2c,
Andrea
--
Andrea Bedini
https://www.andreabedini.com
_______________________________________________
ghc-devs mailing list
ghc-devs@haskell.org
http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs
-- With regards, Jappie Klooster e: hi@jappie.me w: https://jappie.me m: +31644237437