Values from command-line switch -s are not in UTF-8

$ perl -CASD -e'
   print "$^V\n"
      , map { "$_ ".length."\n" } $a, $ARGV[0]
   ' -s -- -a=ก ข
v5.40.2
à¸ 3   # expect: ก 1
ข 1

-s, wow. It's been a hot minute...

-s is discouraged in the docs and is extremely rudimentary. This could be considered a bug since -CA should interpret @ARGV in the internal upgraded encoding, but apparently that doesn't apply before -s. I would recommend using Getopt::Long instead, or an option parser that may be more oneliner friendly like https://metacpan.org/pod/Opt::Imistic.

I also recommend https://metacpan.org/pod/Decode::ARGV instead of -CA and https://perldoc.perl.org/open or https://metacpan.org/pod/open::layers instead of -CSD in any serious code, because -C does not validate that the input is UTF-8 nor does it restrict its output to UTF-8, but instead assumes everything is in Perl's internal upgraded encoding which is a superset of UTF-8.

I think -s is good enough for one-liner code (like awk -v), so please fix it to be consistent with @ARGV.

To ensure backward compatibility, s may be added to be a value for -C (e.g. -CsASD).

Also, it would be great to have a hash variable (specified by -s:) containing all -s variables, e.g.:

$ perl -E'
   say join ", ", map { "$_=$s{$_}" } keys %s
   ' -s:s -- -a=1 -b=2
a=1, b=2

mark -s switches as UTF-8 with -CA

Navigation Menu

Repository navigation

Values from command-line switch `-s` are not in UTF-8 #23377

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Values from command-line switch -s are not in UTF-8 #23377

Description

Activity

guest20 commented on Jun 19, 2025

Grinnz commented on Jun 19, 2025

agguser commented on Jun 19, 2025

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

Values from command-line switch `-s` are not in UTF-8 #23377