Hi, 1. Have you seen the "prettiest printer" article here? https://jyp.github.io/posts/towards-the-prettiest-printer.html It says:
Wadler’s design fares somewhat better. It does not suffer from the above problem… /by default/. That is, it lacks the capability to express that sub-documents should be vertically aligned — compositionally.
...
*
Objection 2: /Leijen’s extension of Wadler’s design solves the issue: it provides an |align| combinator./
A package based on (a later version of) the design in this article is available here: https://hackage.haskell.org/package/pretty-compact This claims to be more ideal ("Prettiest") than either the Hughes ("Pretty") or Wadler ("Prettier") printers. I think it uses dynamic programming to avoid being too slow. If I understand correctly, GHC internally uses a version of the Hughes pretty printer, not the Wadler-Leijen one. 2. Doesn't the wl-print package already have a `nest` combinator? https://hackage.haskell.org/package/wl-pprint-1.2.1/docs/Text-PrettyPrint-Le... It also has the `align` combinator. If I remember correctly, these are part of the Leijen extension to Wadler. Are these not enough to get the behavior that you want? 3. Have you seen hindent? It has a module called HIndent.Pretty that might be relevant to laying out Haskell source. Does that help? -BenRI On 4/17/23 6:50 AM, Johannes Waldmann wrote:
Dear Cafe,
I was looking for a way to pretty-print Haskell literals (with lists, tuples, records with named and positional notation) like this example
( Leftist { tree = Branch { left = Branch { left = Leaf, key = 4, right = Leaf } , key = 3 , right = Leaf } , refs = listToFM [ ( Ref 13, [ 0 ] ), ( Ref 17, [ ] ) ] } , [ Any, Any ] )
for each sub-structure, the indentation level (for the following lines) should increase - by a _fixed_ amount. in the above example: line break after "tree = Branch". But (missing from this example), line break _before_ the list starts in "{ foo = [ 42 , ... ] ... }".
I found this impossible to do with wl-pprint but perhaps I did not try hard enough.
Instead, I "invented" combinators `nest` and `skip` and made this prototypical implementation https://gitlab.imn.htwk-leipzig.de/autotool/all0/-/blob/master/todoc/src/Tex... (it has some explanatory text at the top) see also https://gitlab.imn.htwk-leipzig.de/autotool/all0/-/issues/960
but certainly this cannot be a new idea.
While I do like the semantics (in the context of my application), I don't like the performance of my implementation. What am I doing wrong? It's just updating indentation level and current position, this should not take any time at all?
Of course, it would be best if I don't need the implementation at all - if the effect could be achieved via some combinators in established libraries (that have optimized implementation).
Any pointers appreciated.
Best regards - J. _______________________________________________ Haskell-Cafe mailing list To (un)subscribe, modify options or view archives go to: http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe Only members subscribed via the mailman list are allowed to post.