Excessive String Creation

[stevenjbaldwin]

When using the SOM_LEFTMOST expression flag, all matching strings are returned within the returned list of match objects. For large strings (1+MB) and certain regexps this could cause orders of magnitude more strings to be created. If the scan method could return these strings as a lazy collection such as a Stream instead of a List, this could avoid consuming too much heap memory via the string pool. Alternatively, just the start and end offsets could be returned so the caller can control the number of strings created.

[gliwka]

Interesting use case. Returning a stream definitely makes sense for large amount of matches. I'll take a stab at implementing it. Will actually need this soon. too.