docs(tvix/eval): propose builtin "inlining" optimisation

Change-Id: I96a187792a1fd48cffd6b56ec22347aee8cae3af
Reviewed-on: https://cl.tvl.fyi/c/depot/+/6526
Autosubmit: sterni <sternenseemann@systemli.org>
Reviewed-by: tazjin <tazjin@tvl.su>
Tested-by: BuildkiteCI
This commit is contained in:
sterni 2022-09-06 21:39:03 +02:00 committed by clbot
parent cbde0292b6
commit b6ab02e48b

View file

@ -62,6 +62,24 @@ optimisations, but note the most important ones here.
The same thing goes for resolving `with builtins;`, which should The same thing goes for resolving `with builtins;`, which should
definitely resolve statically. definitely resolve statically.
* Inline fully applied builtins with equivalent operators [medium]
Some `builtins` have equivalent operators, e.g. `builtins.add`
corresponds to the `+` operator, `builtins.hasAttr` to the `?`
operator etc. These operators additionally compile to a primitive
VM opcode, so they should be just as cheap (if not cheaper) as
a builtin application.
In case the compiler encounters a fully applied builtin (i.e.
no currying is occurring) and the `builtins` global is unpoisoned,
it could compile the equivalent operator bytecode instead: For
example, `builtins.add 20 22` would be compiled as `20 + 22`.
This would ensure that equivalent `builtins` can also benefit
from special optimisations we may implement for certain operators
(in the absence of currying). E.g. we could optimise access
to the `builtins` attribute set which a call to
`builtins.getAttr "foo" builtins` should also profit from.
* Avoid nested `VM::run` calls [hard] * Avoid nested `VM::run` calls [hard]
Currently when encountering Nix-native callables (thunks, closures) Currently when encountering Nix-native callables (thunks, closures)