use re '/aansx'; breaks split " ", ...

[druud]

$ perldoc -f split
[...] As another special case, "split" emulates the default behavior of the command line tool awk when the PATTERN is either omitted or a string composed of a single space character
[...] In this case, any leading whitespace in EXPR is removed before splitting occurs, and the PATTERN is instead treated as if it were "/\s+/";
[...] in Perl 5.18.0 and later this special case is triggered by any expression which evaluates to the simple string " ".
[...] If omitted, PATTERN defaults to a single space, " ", triggering the previously described awk emulation.
[...]

$ perl -Mstrict -wE'

  my $s= "  foo  bar  ";

  say for split " ", $s;

  { use re qw( /aansx );
    say for split " ", $s;
  }
'

foo
bar
 
 
f
o
o
 
 
b
a
r 

---
Site configuration information for perl 5.24.3:

Configured by root at Tue Feb 13 14:35:15 CET 2018.
[...]
Locally applied patches:
    fix for #132063
    fix for #132227
[...]

[druud]

I guess that split needs (?^: ... ).

[richardleach]

Simplifies to:

perl -Mstrict -wE'

  my $s= "  foo  bar  ";
  say for split / /, $s;
  say for split / /aansx, $s;
'

I'm confused as to what the bug/issue is here. Surely the use of x in the use re statement tell the interpreter to treat the subsequent line as if it had been written say for split "", $s;? In which case the output is what should be expected.

Is this perhaps somewhere where we could usefully improve the documentation, most likely that of the re module?

[mauke]

split " " is not the same thing as split / /. (Also, this is still broken in blead.)

[richardleach]

split " " is not the same thing as split / /. (Also, this is still broken in blead.)

Doubtless I will look at this again in the morning and feel really stupid, but how do they differ?

They both generate the same op tree with the same split op:

9              </> split(/" "/)[t2] lKM/IMPLIM ->a

and the same final regex program:

Final program:
   1: EXACT < > (3)
   3: END (0)

Or is the bug that they do generate the same thing but in fact should not?

[mauke]

$ perl -wE 'say "[$_]" for split / /, "  foo\tbar"'
[]
[]
[foo    bar]
$ perl -wE 'say "[$_]" for split " ", "  foo\tbar"'
[foo]
[bar]
$

As for how this is implemented, I think it involves regex flags (RXf_WHITE and RXf_SKIPWHITE, which perlreapi claims is unused, but pp.c still checks for it (?)).

Anyway, as the documentation quoted by OP explains, passing a string consisting of a single space to split is supposed to act like a flag that makes split do the equivalent of $string =~ /\S+/g as a special case. It is not treated as a regex, so any use re flags should be ignored.