RE: LocationTemplate-1G and 2G results

Doesn't it seem that we are going too far? The idea of someone including
an '=' in a name seems like the kind of thing we should forbid outright,
not code our way around.

Tony Rogers
tony.rogers@ca.com

-----Original Message-----
From: www-ws-desc-request@w3.org [mailto:www-ws-desc-request@w3.org] On
Behalf Of Jonathan Marsh
Sent: Saturday, 17 March 2007 10:01
To: 'Youenn Fablet'
Cc: 'keith chapman'; www-ws-desc@w3.org
Subject: RE: LocationTemplate-1G and 2G results


I think "=" could harm a query param.  "?{name}={value}" when name has
an '=' will result in an ambiguity between where the name ends and the
parameter begins if "=" isn't encoded.

Not the worst problem I can imagine though...

Jonathan Marsh - http://www.wso2.com -
http://auburnmarshes.spaces.live.com
 

> -----Original Message-----
> From: Youenn Fablet [mailto:youenn.fablet@crf.canon.fr]
> Sent: Friday, March 16, 2007 6:18 AM
> To: Jonathan Marsh
> Cc: 'keith chapman'; www-ws-desc@w3.org
> Subject: Re: LocationTemplate-1G and 2G results
> 
> Thanks for the input.
> See inline.
> Regards,
>     Youenn
> 
> Jonathan Marsh wrote:
> > Youenn,
> >
> > I'm updating the expected results, but there are still a few
> questionable
> > items which I believe are errors on your end, which affect all three
> test
> > cases.
> >
> > 1) The character #xA3 doesn't seem to be encoded correctly.  Per the
> rules
> > in the IRI spec, UTF-8 encoding for characters up to #x7f is 
> > trivial,
> but
> > for characters over #x80 it gets more complicated.  That's why 
> > included
> a
> > character over #x80 ;-).  My read is that #xA3 gets percent-escaped 
> > according to these rules as %C2%A3.
> >
> Yes, you are right.
> I will integrate the whole 1 to 4 bytes UTF8 conversion before doing 
> the % encoding.
> > 2) There is an extra ?, encoded as %3F, following the "path"
parameter.
> > There isn't supposed to be one in the test data.
> >
> ok, I will remove the ? at the end of the path parameter value.
> > 3) You also encode "=" even though there isn't a MUST or SHOULD
> requirement
> > in the spec to do so.  You conform, but it means I'll have to adjust

> > the expected value in the test suite to represent the MUSTs + the 
> > SHOULDs +
> "=".
> > I'm happy enough to do that, but just thought I'd point it out in 
> > case
> you
> > think there's an erratum there.
> >
> Reading the current html draft, '=' does not appear in any of the 
> characters lists.
> This implies that one must encode it, which may not be the intention.
> It seems to me that '=', as a query parameter value character,  is not

> harmful to the query parsing.
> I do not know if '=' has a specific meaning as a path character.
> It might however be more coherent to add '=' to the SHOULD list.
> 
> 
> > Jonathan Marsh - http://www.wso2.com -
> http://auburnmarshes.spaces.live.com
> >

Received on Sunday, 18 March 2007 23:27:14 UTC