From: Chris Vine <vine35792468@gmail.com>
To: Sam Halliday <sam.halliday@gmail.com>
Cc: guile-user@gnu.org
Subject: Re: emulate "sum type" pattern matching?
Date: Thu, 12 Mar 2020 14:46:22 +0000 [thread overview]
Message-ID: <20200312144622.0c660172af70ced0493ce9b1@gmail.com> (raw)
In-Reply-To: <87sgier7eb.fsf@gmail.com>
On Wed, 11 Mar 2020 19:58:04 +0000
Sam Halliday <sam.halliday@gmail.com> wrote:
> Hi all,
>
> I have read the Guile manual as my introduction to Guile. I am very
> impressed at how mature this project is and was overwhelmed by the
> feature set, which seems to be on-par with heavily invested technologies
> like Java and the JVM.
>
> I am considering using Guile for a project because I love Emacs lisp and
> know it very well. Emacs lisp has some limitations that I feel Guile
> overcomes, e.g. multithreading, a superior regexp engine, a module
> system, and parsers.
>
> However, there is one feature that is critical to the development of the
> project and I was hoping to be able to implement it through a macro: sum
> type pattern matching.
>
> By that, I mean in the sense of Haskell sum types, which I understand
> are similar to C++ union types. Roughly translated into GOOP, and using
> Scala's encoding of sum types, this would look like record types that
> are all children of a superclass. I noticed that Guile has support for
> discovering all direct subclasses at runtime, but is this facility
> available at compiletime?
>
> An example of how I would want to use this feature can be described in
> terms of the XML calculator in the Guile Manual.
> https://www.gnu.org/software/guile/manual/html_node/sxml_002dmatch.html#Catamorphisms
> which looks like
>
> (define simple-eval
> (lambda (x)
> (sxml-match x
> [,i (guard (integer? i)) i]
> [(plus ,x ,y) (+ (simple-eval x) (simple-eval y))]
> [(times ,x ,y) (* (simple-eval x) (simple-eval y))]
> [(minus ,x ,y) (- (simple-eval x) (simple-eval y))]
> [(div ,x ,y) (/ (simple-eval x) (simple-eval y))]
> [,otherwise (error "simple-eval: invalid expression" x)])))
>
> If the sxml-match was aware that it was matching over a superclass of
> plus, minus, times, div then the "otherwise" line would be redundant and
> (most importantly) if I were to forget to match over one of the
> subclasses I would get a compiler error.
>
> And that's basically my usecase in a nutshell: exhaustive pattern
> matching over a tree-like structure (an AST, in fact). But I'll have
> lots of different trees so I don't want to have to manually write a
> pattern match macro every time I define a "sum type"... although that
> said I do have some ideas on how to abstract that. But I don't really
> want to go down a lisp macro rabbit hole at the very beginning...
guile's built-in pattern matcher (ice-9 match) enables you to match on
symbols, literals, pairs, lists, vectors and records, but I don't think
it enables you to match on GOOPS objects - someone may contradict me on
that, but at least I have never tried doing so (I don't like GOOPS nad
I rarely use it). The other problem is that discriminated unions are a
better fit with statically typed languages such as Haskell and the MLs;
in dynamically typed languages there is a sense in which every variable
name represents a union of unlimited extent but which enables its
current type to be interrogated.
next prev parent reply other threads:[~2020-03-12 14:46 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-03-11 19:58 emulate "sum type" pattern matching? Sam Halliday
2020-03-12 14:46 ` Chris Vine [this message]
2020-03-12 14:56 ` Ricardo Wurmus
2020-03-12 17:16 ` Chris Vine
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
List information: https://www.gnu.org/software/guile/
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20200312144622.0c660172af70ced0493ce9b1@gmail.com \
--to=vine35792468@gmail.com \
--cc=guile-user@gnu.org \
--cc=sam.halliday@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).