zorba-coders team mailing list archive

[Juan Zacarias <juan457@xxxxxxxxx>]

I changed the implementation as you suggested.

> Unless I'm mistaken, there's no requirement that a fragment identifier must
> follow a forward-slash. The following is a valid URI with a fragment
> identifier:
> 
>   http://www.example.com/foobar#zot
> 
> So the check for (lNormUri.at(found-1) == '/') is incorrect. Also I think it
> might cause a crash if the URI is the string "#" (which is probably not valid
> anyway, but still we shouldn't crash).
> 
> More generally: I just spent a while reading RFC 3986 to see whether a simple
> character-search for '#' was sufficient to identify a fragment, when it
> occurred to me that I shouldn't be trying to answer that question manually. I
> think a better solution here would be to construct a zorba::URI object
> (src/zorbatypes/URI.h) from lNormUri and then call get_encoded_fragment() on
> it to see if there is any fragment. I don't like introducing that much
> overhead to perform this function, but parsing URIs is tricky and IMHO it
> makes sense to consolidate all of that functionality in a single place, namely
> the URI class.
-- 
https://code.launchpad.net/~zorba-coders/zorba/bug1073091/+merge/139346
Your team Zorba Coders is subscribed to branch lp:zorba.