Hi,...
additional comment about the performance.
It seams that we need to write our own handwritten parser for the URI stuff. What do you think?
Any specialized parser would be faster than generic parser such as regexp. And, regexp library is not optimized for matching long strings (together with some regexp constructs) due to its recursive design. See regexp bugzilla for more details.
Thus, I would prefer not to use regexp in tight spots, but use something simpler & faster.
Vadim
(disclaimer: I've got no idea how regexp got into sourceresolving in the first place)
