unofficial mirror of bug-guile@gnu.org 
 help / color / mirror / Atom feed
* bug#49707: Documentation and behavior differ for match (not ...) pattern
@ 2021-07-23  9:45 Jakub Wojciech
  0 siblings, 0 replies; only message in thread
From: Jakub Wojciech @ 2021-07-23  9:45 UTC (permalink / raw)
  To: 49707

[-- Attachment #1: Type: text/plain, Size: 1643 bytes --]

The documentation states:
> (not pat_1 ... pat_n)           if all pat_1 thru pat_n don't match

The code only implements (not pat), for a singular pattern, e.g:
  (match 2
    ((not 1) 'not-one)
    (1 'one)
    (2 'two))
  => not-one

According to the documentation this should work, but the result is erroneous:
  (match 3
    ((not 1 2) 'not-one-nor-two)
    (1 'one)
    (2 'two)
    (3 'three))
  => three

So it fails silently.

RhodiumToad on #guile proposed the simple fix that I took a liberty of
attaching to this message.
It adds a clause for (not ...), delegating it to 'or': (not (or ...)).

However RhodiumToad also raised another issue: is the code wrong or is
the documentation wrong?

The documentation in the file itself states:
> The 'not' operator succeeds if the given pattern doesn't match.
The test from upstream also only checks for the singular pattern inside
the 'not' clause.
This means that the idea behind this code is to allow one and only one
pattern.

Although I lean towards fixing the code to match the Guile's
documentation (i.e. applying the attached patch), I also wonder about
the relation with the upstream - Chibi Scheme.
There are three possibilities:
1. Diverge from their implementation.
2. Try to convince them to apply that patch too.
3. Make passing more than one pattern to 'not' clause a syntax error and
   changing the info manual documentation.

The question is: which one do we want to choose?

The rationale for not selecting option 3 is the fact that the change is
non-breaking, adds a functionality, and conforms to both SRFI-200 and
SRFI-204 drafts and the original Wright-Duba paper.


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: match-patch.diff --]
[-- Type: text/x-patch, Size: 1009 bytes --]

diff --git a/module/ice-9/match.upstream.scm b/module/ice-9/match.upstream.scm
index b1fc371b8..f12981cb3 100644
--- a/module/ice-9/match.upstream.scm
+++ b/module/ice-9/match.upstream.scm
@@ -115,7 +115,7 @@
 ;;> @example{(match 1 ((or x) x))}
 ;;> @example{(match 1 ((or x 2) x))}
 
-;;> The @scheme{not} operator succeeds if the given pattern doesn't
+;;> The @scheme{not} operator succeeds if none of the given patterns
 ;;> match.  None of the identifiers used are available in the body.
 
 ;;> @example{(match 1 ((not 2) #t))}
@@ -355,6 +355,8 @@
      (match-extract-vars (or p ...) (match-gen-or v (p ...) g+s sk fk i) i ()))
     ((match-two v (not p) g+s (sk ...) fk i)
      (match-one v p g+s (match-drop-ids fk) (sk ... i) i))
+    ((match-two v (not p ...) g+s (sk ...) fk i)
+     (match-two v (not (or p ...)) g+s (sk ...) fk i))
     ((match-two v (get! getter) (g s) (sk ...) fk i)
      (let ((getter (lambda () g))) (sk ... i)))
     ((match-two v (set! setter) (g (s ...)) (sk ...) fk i)

^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2021-07-23  9:45 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-07-23  9:45 bug#49707: Documentation and behavior differ for match (not ...) pattern Jakub Wojciech

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 NNTP newsgroup(s).