completely on board for a type-centric path there,...
# development
r
completely on board for a type-centric path there, but: 1) what could that look like 2) do we really need it
w
...right. bit too early to know precisely.
r
I think so too - we can iterate on that when we need it.
so maybe we even punt on weighting for now
w
yes.
imo.
r
👍
w
so... we... sortof know what it would look like. i tried it out during the engine experiments
r
oh?
w
(or John did... it's blurry now =P)
now gone, because i removed the "goal" concept from the scheduler
and to be clear: i support the idea that a
@console_rule
should not have any output. the idea of "defining them in terms of their product requirements" is very much a thing
would just love to never let them depend on one another.
r
interesting. yeah, agree re: no knowledge of each other
maybe theres a 1:1 mapping of goal ->
@console_rule
to start to enforce that?
w
yes, i would really love to start there.
if we can pull that off, it would make me very happy.
r
sounds great to me
w
because we must allow type-level extension (cc @aloof-angle-91616) to allow for pluggability of rules anyway
ie: "i can install another rule that provides TestResults" or "i can install another rule that provides Classpaths", etc