This is to do with URL processing it appears to be isolated to image tag
URL parsing (doesn’t seem to be affected by normal URLs).
It trims the any number dots before the first / in relative URL’s. I
kind of depend on this as the code I’m generating is part of a
publishing system that needs to be able to handle relative links so
pages can be distributed offline.
On a side note, congrats Jason on the rewrite! Will it be long before
this new version becomes standard?