Cannot use case+else inside match+else

[Shu-Hung You]

When using a case expression with an else clause in inside a match
expression, the expander complains about case having a bad syntax.
However, the use of case expressions outside of match are fine. Is
there anyway to get around this?

Currently, I just replace it with another match.

#lang racket/base

(require racket/match)

(match 6
[else
(case 5
[else 7])])

#|
unsaved editor:8:6: case: bad syntax (not a datum sequence) at: else
in: (case 5 (else 7))
#(78 4)
|#

[John Clements]

I may be missing something obvious, here, but why are you using the name “else” to bind the result in the match clause? I usually use the name “other”, which might solve your problem.

John

> --
> You received this message because you are subscribed to the Google Groups "Racket Users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to racket-users...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

[Sam Tobin-Hochstadt]

This is because 'case' expects a particular binding for 'else', but you shadowed it with 'match', which does not treat 'else' differently than any other identifier. 

Sam

[Shu-Hung You]

Ah!

Thank you for pointing this out. I've always thought that `else' is
treated as a keyword in `match' just as what `cond' does. That
explains everything.

[Sorawee Porncharoenwase]

I usually use _ for the otherwise case because I usually match on an identifier, so I have an id to refer to the matched value already if I need to. The only case I don’t use _ is when I compute something inside the match and also need to use its result:

(match (foo bar)
  [... ...]
  [result-of-foo-bar ...])

[Robby Findler]

I have been bitten by this one too, it is a bit pernicious. Perhaps we
can adjust the documentation to make more clear that `else` isn't
special in `match`?

Robby

[Greg Hendershott]

Yep, I also spent a non-zero number of years not even realizing "else"
wasn't a magic literal for match. And then remembering to use _
instead, as Sorawee suggested.


I think this shows why it's usually better for syntax to use
#:keywords instead of literals? If cond and case used #:else, this
wouldn't be a problem.

(Racket inherited those from Scheme. But we don't have to follow that
example in our own, new macros.)

[Jon Zeppieri]

Previously on this program... 

https://lists.racket-lang.org/dev/archive/2013-May/thread.html 

[racket-dev] else clauses: possible change to match?

[Ben Greenman]

Here's a suggestion for the docs:

https://github.com/racket/racket/pull/2505

[Laurent]

Wait, that means that in an interactive session, if you ever happen to redefine `else', you can't use `case' anymore?

On Tue, Feb 26, 2019 at 5:03 AM Ben Greenman <benjamin...@gmail.com> wrote:

Here's a suggestion for the docs:

https://github.com/racket/racket/pull/2505

[Jens Axel Søgaard]

I suppose you could (re)require it again. 

--

--
Jens Axel Søgaard

[Laurent]

Good point. I wasn't sure that would work---it does.

[Shu-Hung You]

And because it is nothing else but a usual binding, it's possible to
prefix-in or rename-in as in

(require (prefix-in r: racket/base))
(case 5 [r:else 'ok])