Post #1688995
2026-03-30 15:32 UTC
Replies (3)
-
@tim_salabim@fosstodon.org 2026-03-30 16:55
@defuneste @lwpembleton @drdcarpenter @mschmidty @milesmcbain out of interest, why the reservation about yaml?
-
@danwwilson@rstats.me 2026-03-30 21:47
@defuneste @lwpembleton @drdcarpenter @mschmidty @milesmcbain We do a list type thing in {targets}and then have a targets factory that brings each list element into its own target. It works pretty well for our purposes and means we’re able to save our. Config as an rds object if we want to.
-
@milesmcbain@fosstodon.org 2026-04-01 12:23
@defuneste @lwpembleton @drdcarpenter @mschmidty as you know, I have called this an anti-pattern. You’re putting analysis behind a single declarative interface with many parameters. This type of interface is brittle because the combinatoric explosion of possible parameter values makes it difficult to create appropriate behaviours for all combinations. The example was only small, yet I think still failed to handle a case where not every selected trait has a defined weight(?).