Re: [apps-discuss] Parsing text into URLs that doesn't conform to RFC 3986

[<[e-mail]>]

What really concerns me about the apparent objective of making all URL parsers produce identical results from the set of tests produced by Whatwg, is that it seems to ignore the fact that URLs are created and consumed in very different contexts.




When text is entered in the address bar of a web browser and it is parsed as a URL, then I understand the desire to be as tolerant as possible of errors.  I see minimal danger in making an educated guess as to what the user intended because the browser will likely correct the error and show the user what the corrected URL is, before making the safe request.




However, when I use the System.Uri class to parse some text that I find in a HTTP response or configuration file, it would really concern me if the parser started making significant assumptions about how to generate valid URLs from text that doesn’t conform to the syntax of RFC 3986.




I would be really interested in seeing the development of a specification that describes the optimum way of converting a string that doesn't conform to RFC 3986 syntax into one that does.  That I can get behind.  Changing .NET's System.Uri to make best guesses, I can't.




Darrel