diff options
Diffstat (limited to 'docs/wiki_src/coding')
| -rw-r--r-- | docs/wiki_src/coding/compilation.md | 23 | ||||
| -rw-r--r-- | docs/wiki_src/coding/data-structures.md | 2 | ||||
| -rw-r--r-- | docs/wiki_src/coding/meta-programming.md | 10 |
3 files changed, 17 insertions, 18 deletions
diff --git a/docs/wiki_src/coding/compilation.md b/docs/wiki_src/coding/compilation.md index 9863c21..330bc9f 100644 --- a/docs/wiki_src/coding/compilation.md +++ b/docs/wiki_src/coding/compilation.md @@ -1,6 +1,6 @@ # Compilation -Bruijn can be compiled to John Tromp’s binary lambda calculus (BLC). +Bruijn can be compiled to John Tromp's binary lambda calculus (BLC). BLC uses the following encoding: @@ -12,26 +12,27 @@ BLC uses the following encoding: There are two modes of compilation: -- **Bitwise** compiles to BLC and encodes every bit as 1 bit and pads - the last remaining byte: `bruijn -b path` -- **ASCII** compiles to BLC and encodes every bit as 1 ASCII character - (`'0'`/`'1'`): `bruijn -B path` +- **Bitwise** compiles to BLC and encodes every bit as 1 bit and pads + the last remaining byte: `bruijn -b path` +- **ASCII** compiles to BLC and encodes every bit as 1 ASCII character + (`'0'`/`'1'`): `bruijn -B path` ## Compilation overhead Typical compilation to BLC results in much redundant code, since every used function gets substituted and translated separately. In -`((+3) + (+4) + (+3))`, for example, `add` gets compiled to BLC two -times, resulting in a redundant overhead of around 3500 bits. +`((+3) + (+4) + (+3))`{.bruijn}, for example, `add`{.bruijn} gets +compiled to BLC two times, resulting in a redundant overhead of around +3500 bits. This is because BLC was never intended for compilation of normal programs, but mainly as an academic encoding model. This also means that -it’s quite good for writing very expressive and minimal programs -(i.e. obfuscated code golfing, see [John Tromp’s +it's quite good for writing very expressive and minimal programs +(i.e. obfuscated code golfing, see [John Tromp's IOCCC](https://ioccc.org/2012/tromp/hint.html)). -Most programs, however, won’t be golfed and can result in rather large -compiled programs. While there’s not really any practical need for +Most programs, however, won't be golfed and can result in rather large +compiled programs. While there's not really any practical need for compilation aside from golfing, you could still use the [BLoC](https://github.com/marvinborner/bloc) project to optimize redundant terms. diff --git a/docs/wiki_src/coding/data-structures.md b/docs/wiki_src/coding/data-structures.md index 0516bf7..3d2a387 100644 --- a/docs/wiki_src/coding/data-structures.md +++ b/docs/wiki_src/coding/data-structures.md @@ -39,7 +39,7 @@ syntactic sugar. - default syntactic sugar for numbers (optional suffix `t`), e.g. `(+3)`{.bruijn} -- encoding: `(+4)`{.bruijn}=`[[[[(1 (1 3))]]]]`, +- encoding: `(+4)`{.bruijn}=`[[[[(1 (1 3))]]]]`{.bruijn}, `(-4)`{.bruijn}=`[[[[(2 (2 3))]]]]`{.bruijn} - faster and more compact than binary[^1] - positive and negative numbers diff --git a/docs/wiki_src/coding/meta-programming.md b/docs/wiki_src/coding/meta-programming.md index 5d1caa0..334e285 100644 --- a/docs/wiki_src/coding/meta-programming.md +++ b/docs/wiki_src/coding/meta-programming.md @@ -1,4 +1,4 @@ -# Meta programming +# Metaprogramming Bruijn has a homoiconic meta encoding inspired by Lisp's quoting feature. @@ -8,11 +8,9 @@ self-interpretation](https://text.marvinborner.de/2023-09-03-21.html). ## Encoding -``` bruijn -`X ⤳ [[[2 (+Xu)]]] -`(M N) ⤳ [[[1 `M `N]]] -`[M] ⤳ [[[0 `M]]] -``` + `X ⤳ [[[2 (+Xu)]]] + `(M N) ⤳ [[[1 `M `N]]] + `[M] ⤳ [[[0 `M]]] Any quoted term gets converted to this encoding: |