Re: [pcre-dev] Limiting the Unicode validity check to the m…

>so the check could start at the offset minus the maximum lookbehind.

Do you mean working thru UTF* backwards (for as many characters as
lookbehind want) to compute the offset?

But doesn't that imply checking twice, since there can be invalid UTF*
in the lookbehind part?

Or did you mean that this feature would perform a backward only check
first (for the lookbehind) then perform regular forward checking while
eating the subject starting from the (unchanged) offset?

Just curious.