On Friday, 29 January 2016 at 17:49:58 UTC, Nick Sabalausky wrote:
I don't recall: Does that parse the source for comments on its own or does it still use dmd's json (or html) output?
Does it on its own. (Well, except the search results page, it still uses the json, but I'm fixing that soon and the main body pages already do their own thing.)
Brian Schott's libdparse does the bulk of the work, independently of dmd. A big reason for this is that doing changes on dmd is a pain in the butt, and another one is that dmd is optimized toward compiling code (as it should be!) which isn't always ideal for doc generation (like version(Windows) docs being left out if you happen to be on a Linux box.)
So doing it myself frees me from dmd's design constraints as well as dmd's development process.