W3C home > Mailing lists > Public > public-qa-dev@w3.org > February 2009

Re: linkchecker's Makefile.PL and CPAN trick

From: olivier Thereaux <ot@w3.org>
Date: Mon, 2 Feb 2009 14:29:17 -0500
Cc: public-qa-dev@w3.org
Message-Id: <3459DB5E-F79B-4704-AFA4-73329DD839BC@w3.org>
To: Ville Skyttä <ville.skytta@iki.fi>

Hi Ville,

On 31-Jan-09, at 10:56 AM, Ville Skyttä wrote:
> The only thing I'm a bit uncertain of is whether application/ 
> vnd.wap.xhtml+xml
> should be kept in the Accept header.  Even though it currently in  
> the CVS
> version has a q value of 0.6 (lower than everything except */*),  
> some large
> sites such as Google and Yahoo now decide to serve mobile versions  
> of their
> sites to the link checker.

This is certainly part of a disturbing trend that many sites show,  
obviously misimplementing content negotiation. In this case I share  
your uncertainty, but would be tempted to:
1) keep the link checker as it is now. I somewhat hope that sites  
doing that kind of on-the-fly content adaptation will have the same  
URI mapping for "web" and "mobile"
1) send a quick mail to the big sites identified
2) write about the issue on a W3C blog, try to put it in a way that is  
not accusatory but wonder what are the causes of this implementation  
(bad mobile User-Agents issuing bogus Accept strings?) and what can be  
made to keep negotiation relevant.

>
>> Would you prefer doing it or should I?
>
> I can take care of it, let's say earlyish next week.  The only thing  
> in
> addition to possibly the above I have on the TODO list is
> mentioning "perl -MCPAN -e 'install W3C::LinkChecker'" in the docs.

Good point. I have added a bit of text and see that you've made  
improvements to that draft.

Thanks!
-- 
olivier
Received on Monday, 2 February 2009 19:29:29 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 19 August 2010 18:12:49 GMT