The \b is tricky, as I can use it to get three different results from my three subroutines:

# Add this to the list of cases:

    [ "a\b\\b" => '\\/',  'xy',  "a\bxb"]

# Results:

not ok 22 - simulate
#   Failed test 'simulate'
#   at ./2.pl line 64.
#          got: 'xb'
#     expected: 'xb'
ok 23 - cheat
Replacement list is longer than search list at (eval 45) line 1.
not ok 24 - use
#   Failed test 'use'
#   at ./2.pl line 66.
#          got: '\b'
#     expected: 'xb'
[download]

Switching to Test2::V0 explains the test 22:

not ok 22 - simulate
# Failed test 'simulate'
# at ./2.pl line 64.
# +-------+----+-------+
# | GOT   | OP | CHECK |
# +-------+----+-------+
# | a\bxb | eq | xb    |
# +-------+----+-------+
[download]

Sorry I'm too tired now, I'll check tomorrow.

Multiple levels of escaping are often kind of brainf*ck.

> would expect the strings coming in to contain the characters to operate on, not strings that Perl can turn into the characters to operate on by evaluating them.

this escape_metas can also be used with a character class [METAS] for with multiple metas.

> The \b is tricky, as I can use it to get three different results from my three subroutines:

exactly, as I said, it depends what the intended behavior is.

FWIW

my intention was to implement a sub tri = "transliterate interpolated" which allowed to write tri($input,$left,$right) and implemented $input =~ tr/LEFT/RIGHT/ while sticking close to tr's behaviour but without breaking.

I'll try again tomorrow. after GPW. :)

Cheers Rolf
_{(addicted to the Perl Programming Language :)
Wikisyntax for the Monastery}

> my intention was to implement a sub tri = "transliterate interpolated" which allowed to write tri($input,$left,$right) and implemented $input =~ tr/LEFT/RIGHT/ while sticking close to tr's behaviour but without breaking.

That's more complicated than I thought because of the special nature of tr's arguments. :(

double-quotes allow interpolation and escape codes, single-quotes don't

  DB<76> p "a\bc"
c
  DB<77> p length "a\bc"
3
  DB<78> p 'a\bc'
a\bc
  DB<79> p length 'a\bc'
4
  DB<80> p 'a\b' eq 'a\\b'
1
[download]

But tr forbids interpolation while allowing escape codes

  DB<80> say '$l' =~ tr/$l/xyz/r    # no interpolation
xy

  DB<81> say "a\b" =~ tr/a\b/xyz/r  # but escape codes
xy
[download]

This makes evaling tr in a predictable way far too complicated.

Just escaping delimiters won't do, if escape codes are involved.

I think it's wise to give it up here.

Cheers Rolf
_{(addicted to the Perl Programming Language :)
Wikisyntax for the Monastery}