| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
|
|
|
|
|
|
wtf
|
|
More details in the linked blog post.
|
|
Originally got the idea from @sgoguen (YC) and was further inspired by
@hbr from one of their blog posts[1].
The current encoding has some holes and bad performance, so I might
switch to a different one in the future.
The wiki and metaprogramming blog post have been updated accordingly.
[1] https://hbr.github.io/Lambda-Calculus/computability/text.html
|
|
In many cases, shared-by-abstraction BLC is more performant (and notably
smaller) than the current output where every term just gets substituted
(and potentially duplicated) directly.
BLoC in combination with BLoCade's shared BLC target optimizes this
automatically by trying to find the terms that would most benefit from
deduplication and abstracting them respectively.
Paging @tromp since we talked about this.
This commit also introduces better argument parsing using
optparse-applicative.
|
|
|
|
|
|
|
|
|
|
css and blog
|
|
|
|
|
