Éric Araujo <mer...@netwok.org> added the comment:

> Is urlparse meant to follow the generic URI RFC?
No, it predates it.

> IMHO, the patch at least should do the equivalent of
> urlparse.uses_fragment.extend(wsschemes)
> so users of urlparse can do the checking for fragment != "", required for 
> ws/wss on their own.
That’s probably the most urllib.parse can do, sadly.

> Alternative: make this bug dependent on fixing urlparse for fragment rules in 
> generic
> URI RFC and don't do anything until then?
I’m not sure we can fix urllib.parse that way, because of backward 
compatibility concerns.  We might change the default handling (i.e. when 
parsing an unknown scheme) to comply with the RFC, but I’d much rather have a 
new, clean module.

> This kind of suggestion has come up before
I recall some discussion on that too.  Maybe we should bring it up again on 
python-dev?  I think I read a discussion from years ago where Guido learned 
that the URI syntax was now generic and that urlparse’s design was obsolete.  
There was also someone else who had a new module (was it Nick?) implementing 
the RFC.  IIRC this module was not discussed for inclusion because urllib 
gained many tests for RFC compliance and was thought Good Enough™.

> There is a number of limitations if want to make the parser generic for any 
> scheme. The
> difficult thing being the parsing behavior and requirements as defined by 
> scheme. 
I don’t understand what you need.  If I get the RFC correctly, the point is 
that parsing rules are the same for any and all schemes, then it’s up to the 
application to refuse some component or do any other scheme-specific handling 
of the components.  But the parsing of a URI into components is the same.

----------

_______________________________________
Python tracker <rep...@bugs.python.org>
<http://bugs.python.org/issue13244>
_______________________________________
_______________________________________________
Python-bugs-list mailing list
Unsubscribe: 
http://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

Reply via email to