From @bacek

Hello.

There is suggestion from Pm for Str.capitalize

+

+=item capitalize
+ [...]
+=cut
+
+ .local int pos, is_ws, is_lc
+ pos = 0
+ goto first_char
+ next_grapheme​:
+ if pos == len goto done
+ is_ws = is_cclass .CCLASS_WHITESPACE, tmps, pos
+ if is_ws goto ws
+ advance​:
+ pos += 1
+ goto next_grapheme
+ ws​:
+ pos += 1
+ first_char​:
+ is_lc = is_cclass .CCLASS_LOWERCASE, tmps, pos
+ unless is_lc goto advance
...

This section of code would be much simpler (and more efficient)
by using the C<find_cclass> and C<find_not_cclass opcodes> instead of
individually examining each character one-at-a-time.

--
Bacek

From @azawawi

Hi,

Here is a re-factored version that passes t/spec tests.
To test, please use​:
./perl6 t/spec/S29-str/capitalize.rakudo

./ahmadz &

Index​: src/builtins/any-str.pir

--- src/builtins/any-str.pir (revision 31549)
+++ src/builtins/any-str.pir (working copy)
@​@​ -42,7 +42,6 @​@​

.sub 'capitalize' :method :multi(_)
  .local string tmps
- .local string fchr
  .local pmc retv
  .local int len

@​@​ -54,27 +53,21 @​@​

  downcase tmps

- .local int pos, is_ws, is_lc
+ .local int pos
+ .local string s1
  pos = 0
- goto first_char
- next_grapheme​:
- if pos == len goto done
- is_ws = is_cclass .CCLASS_WHITESPACE, tmps, pos
- if is_ws goto ws
- advance​:
- pos += 1
- goto next_grapheme
- ws​:
- pos += 1
- first_char​:
- is_lc = is_cclass .CCLASS_LOWERCASE, tmps, pos
- unless is_lc goto advance
- $S1 = substr tmps, pos, 1
- upcase $S1
- substr tmps, pos, 1, $S1
- ## the length may have changed after replacement, so measure it
again
+ next_word​:
+ pos = find_cclass .CCLASS_LOWERCASE, tmps, pos, len
+ s1 = substr tmps, pos, 1
+ upcase s1
+ substr tmps, pos, 1, s1
  len = length tmps
- goto advance
+ pos+=1
+ if pos == len goto done
+ pos = find_not_cclass .CCLASS_LOWERCASE, tmps, pos, len
+ if pos == len goto done
+ goto next_word
+
  done​:
  retv = tmps
  .return (retv)

The RT System itself - Status changed from 'new' to 'open'

From @moritz

On Wed Oct 01 11​:25​:25 2008, ahmadz wrote​:

Hi,

Here is a re-factored version that passes t/spec tests.

Thanks, applied as r22479 (and closing ticket now).

@moritz - Status changed from 'open' to 'resolved'