# \[ANN\] pacomb 1.1

**URL:** <https://discuss.ocaml.org/t/ann-pacomb-1-1/8625>\
**Category:** Ecosystem\
**Tags:** ppx, announce\
**Created:** [October 14, 2021, 8:16am UTC](https://discuss.ocaml.org/t/ann-pacomb-1-1/8625 "2021-10-14T08:16:19Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![craff](https://sea2.discourse-cdn.com/flex020/user_avatar/discuss.ocaml.org/craff/32/3281_2.png) [@craff](https://discuss.ocaml.org/u/craff)\
**Post date:** [October 14, 2021, 8:16am UTC](https://discuss.ocaml.org/t/ann-pacomb-1-1/8625/1 "2021-10-14T08:16:19Z")

</div>

Dear camelers,

I am glad to announce the release 1.1 of pacomb.

defined with the `Grammar` module (or indirectly though a PPX extension) to  
the combinators of the `Combinator` module. The library offers _scanner less_  
parsing, but the `Lex` module provide a notion of _terminals_ and _blanks_  
that give a simple way to write grammars in two phases, as usual.

The main advantage of PaComb and similar solutions, contrary to ocamlyacc, is  
that grammars (compiled or not) are **first class values**. This allows using  
the full power of OCaml for manipulating grammars. For example, this is very  
useful when working with syntax extension mechanisms.

Importantly, the **performances** of PaComb are very good: it is only two to  
five times slower than grammars generated by ocamlyacc, which is a compiler.

Defining languages using the `Grammar` module directly is cumbersome. For that  
reason, PaComb provides a BNF-like PPX syntax extension (enabled using the  
`-ppx pacomb.ppx` compilation flag).

Pacomb also support: self extensible grammars, ambiguous grammars (with merge),  
late rejection of rule via raising exception from action code, priority and others.

Documentation is here [https://raffalli.eu/pacomb/](https://raffalli.eu/pacomb/)  
github: [GitHub - craff/pacomb: A parsing library that compiles grammars to combinators using elimination of left recursion](https://github.com/craff/pacomb)

and it is available via `opam install pacomb`

As teaser, the usual calculator example:

```auto
(* The three levels of priorities *)
type p = Atom | Prod | Sum

let%parser rec
     (* This includes each priority level in the next one *)
     expr p = Atom < Prod < Sum
            (* all other rule are selected by their priority level *)
            ; (p=Atom) (x::FLOAT) => x
            ; (p=Atom) '(' (e::expr Sum) ')' => e
            ; (p=Prod) (x::expr Prod) '*' (y::expr Atom) => x*.y
            ; (p=Prod) (x::expr Prod) '/' (y::expr Atom) => x/.y
            ; (p=Sum ) (x::expr Sum ) '+' (y::expr Prod) => x+.y
            ; (p=Sum ) (x::expr Sum ) '-' (y::expr Prod) => x-.y

```

---

<div class="post-metadata">

**Author:** ![craff](https://sea2.discourse-cdn.com/flex020/user_avatar/discuss.ocaml.org/craff/32/3281_2.png) [@craff](https://discuss.ocaml.org/u/craff)\
**Post date:** [October 14, 2021, 6:18pm UTC](https://discuss.ocaml.org/t/ann-pacomb-1-1/8625/2 "2021-10-14T18:18:40Z")

</div>

Forgot:

Documentation is here [https://raffalli.eu/pacomb/](https://raffalli.eu/pacomb/)  
github: [https://github.com/craff/pacomb](https://github.com/craff/pacomb)

and it is available via `opam install pacomb`

---

<div class="post-metadata">

**Author:** ![cemerick](https://sea2.discourse-cdn.com/flex020/user_avatar/discuss.ocaml.org/cemerick/32/1383_2.png) [@cemerick](https://discuss.ocaml.org/u/cemerick)\
**Post date:** [October 15, 2021, 9:12pm UTC](https://discuss.ocaml.org/t/ann-pacomb-1-1/8625/3 "2021-10-15T21:12:37Z")

</div>

This looks pretty neat, especially the syntax afforded by the ppx. I wonder if you have any notes re: how it compares to [angstrom](https://github.com/inhabitedtype/angstrom/)?

---

<div class="post-metadata">

**Author:** ![craff](https://sea2.discourse-cdn.com/flex020/user_avatar/discuss.ocaml.org/craff/32/3281_2.png) [@craff](https://discuss.ocaml.org/u/craff)\
**Post date:** [October 15, 2021, 11:17pm UTC](https://discuss.ocaml.org/t/ann-pacomb-1-1/8625/4 "2021-10-15T23:17:34Z")

</div>

Thanks for your interest.

Briefly: pacomb (short for french “Pas Comb” = “Not Comb”) compiles BNF to combinators but the combinators are  
not intended to be used directly. The interface of the `Comb` module is probably not optimized for this.  
With pacomb you compile BNF to combinators (at runtime).

The main idea of pacomb it to provide a lot of nice feature with a very good performance:

- full context free grammar in O(N^3) using cache and merge combinators.
- notion of “blank” to eliminate meaningless characters. The blank can be changed during parsing.
- byte position for speed and full position with utf8 column by rescanning the file.
- rejection of rule from the action code
- dependent and extensible grammars (the grammar depends from the beginning of the input)
- parametric grammar
- I probably forgot some.

Pacomb is then well suited for languages that embed other languages, natural languages, but standard application works very well too.

The combinator of pacomb are rather unusual: they use continuation and a scheduler (the combinator does not call itself its continuation, it returns to the scheduler that will call the continuation later to ensure that all alternatives are parsed in parallel) hence pacomb parses the buffer from left to right. Pacomb also uses special combinators for the compilation of left recursive grammars. See `Comb.ml` file.