- From: Anne van Kesteren <annevk@annevk.nl>
- Date: Mon, 13 Oct 2014 16:05:29 +0200
- To: Sam Ruby <rubys@intertwingly.net>
- Cc: whatWG <whatwg@whatwg.org>
On Sun, Oct 12, 2014 at 3:01 PM, Sam Ruby <rubys@intertwingly.net> wrote: > Can you explain in JavaScript terms what the difference is between return > failure and terminate? If you simply terminate new URL("http://test:test/") would no longer throw. (The reason is that the parse algorithm is overloaded. One version parses a string and returns a URL. The other parses a string and modifies a URL in place. The former has a need to return failure, the latter can simply terminate.) >> And parse error error would be more >> like flag a parse error, or append a parse error to a list of parse >> errors. It depends a bit on whether the parser decides to halt on the >> first one or not. > > I don't see anything in the prose that indicates that halting on the first > parse error is an option. It's something I've been meaning to add at some point. Mostly for conformance checkers I suppose. >> Did you have a look at the open bugs by the way? There's a chance the >> parsing algorithm will get rewritten at some point to be a bit more >> functional and less state driven. > > Not yet. I'm still seeing a large set of differences between what I am > producing and what is in urltestdata.txt and need to track down whether the > problems are in my implementation, the spec, or in the test results. > > Once those three are in sync; I'll try to look at the bigger picture. Cool. Sounds great. -- https://annevankesteren.nl/
Received on Monday, 13 October 2014 14:05:58 UTC