> [leo - Fri Oct 07 02:11:28 2005]:
>
> During opcode cleanup the find_word_boundary opcode ceased to exist
> (there was no is_word_boundary).
>
> We probably want to have this as a builtin "character class".
I think we can just deprecate (or omit) find_word_boundary altogether as
an opcode. I don't know the full history of find_word_boundary, but I
speculate it was created in anticipation of being useful for regular
expression matching. However, PGE doesn't use find_word_boundary, nor
do I see anywhere that it'll likely be used.
Someone who needs find_word_boundary can generally get there via the
find_cclass and find_not_cclass ops. In particular:
.local int pos
# ...$S0 is string to search...
$I0 = length $S0
$I1 = find_cclass .CCLASS_WORD, $S0, pos, $I0
$I2 = find_not_cclass .CCLASS_WORD, $S0, pos, $I0
the greater of $I1 and $I2 gives the next word boundary after pos. Of
course, if we already know from previous information that the character
at pos is .CCLASS_WORD (or not .CCLASS_WORD), we can get the next word
boundary with just one call to find_not_cclass (or find_cclass).
So, I vote to just eliminate find_word_boundary altogether.
Pm
