W3C home > Mailing lists > Public > www-validator@w3.org > April 2008

Re: please add Accept header to http request containing application/xhtml+xml

From: Etienne Miret <elimerl@gmail.com>
Date: Mon, 21 Apr 2008 17:32:09 +0200
Message-ID: <480CB379.4050200@gmail.com>
To: Dean Edridge <dean@55.co.nz>
CC: David Dorward <david@dorward.me.uk>, www-validator@w3.org, Alexandre Alapetite <alexandre@alapetite.net>

Hi Dean,

I'm pretty much agree with David on those points.

Dean Edridge a écrit :
> Hi,
> 
> Here's the problem with the W3C Validator as I see it.
> 
> The W3C Validator does not send an Accept header by default.
> IMHO it is not good practise for any web server to send XHTML to a
> user-agent without first knowing that the user-agent accepts XHTML. This
> may seem like an odd concept to some but it is just the way things are
> if you want to use XHTML on the web today. In some situations you may
> get away with sending XHTML to everyone, but such situations are rare. I
> myself, will not send XHTML to a user-agent that does not specifically
> declare that it accepts application/xhtml+xml.
I think we have a real-world versus standards issue here.

Standard-compliant content-negotiation requires you to understand the
lack of "Accept" header as "The user-agent accepts anything", hence, you
should send XHTML to such user-agent. However, in the real-world some UA
won't send a correct accept header, and those UA are typically unable to
handle XHTML, so you *need* hacks. There is no way around it. You need 
some dirty hack if you want to support IE.

I don't think however, that the hack should be in the W3C validator 
(which is precisely about standard compliance) rather than in the Web 
server.

> When they could actually just do this if the validator had an accept
> header.
> <?php
> if (stristr($_SERVER['HTTP_ACCEPT'], "application/xhtml+xml"))
> {
>    $mime = "application/xhtml+xml";
> }
> else
> {
>    $mime = "text/html";
> }
> header("Content-Type: $mime; charset=utf-8");
> ?>
As I said, you don't apply correct content-negotiation here. Although
this might be a good thing to do in the real world, you cannot complain 
if it raises issues with a tools that's precisely about respecting 
standards.

> Part a) Have the W3C Validator send, by default, an appropriate Accept
> header with all the media types that it supports. This could look
> something like this:
> Accept: application/xhtml+xml, application/xml; q=0.5, text/html; q=9,
> text/xml, application/smil, application/smil+xml, image/svg+xml; q=0.3,
> ;*/* q=0.1
I feel this is a good idea. At least, it would do no harm. The issue 
with the current lack of Accept header is not that it means the 
validator doesn't understand XHTML, it is that it means the validator 
can handle plain text, PDF or MS Word documents. Since IE7 sends an 
Accept header which includes "application/msword" (with the maximum 
quality value), doing content-negotiation between a Word and an HTML 
document is not purely theoretical. I don't know if some websites do it, 
but it sure is possible.

I had once an issue, when - for test purpose - I added a plain text 
variant to my home page. Well, for several months the validator just 
wouldn't validate my home page. I solved the issue by telling my Web 
server that in case several variants where acceptable and none preferred 
by the UA, it should send first the XHTML one, then the HTML one, and 
the plain text one only as a last resort.

In that case it was easy, but when doing content-negotiation between an 
MS Word and an HTML document, setting the MS Word document as the 
preferred one is - well - realistic.

But of course, as soon as you do content-negotiation, you are likely to 
be willing to test different variants. In that case you need an option 
to set the accept header rather than a default value. In my case, I had 
plain text, HTML and XHTML. The fact that by default the validator would 
check the plain text variant was a annoying, but usually I used a 
specific URL in order to check the HTML or the XHTML variant.

> Part b) Have an ability for users/webdevelopers to add their own
> media-types that would override the default Accept settings that would
> normally be sent by the validator. This is explained above, but I feel
> that it is only beneficial to maintainers of the website being validated
> or other tech like people, so I do not think this would be a total
> solution on it's own.
I think that being able to specify a custom header will only be useful 
to tech like people anyway (well, you need to know what 
content-negotiation is). But being such a person, I would much more 
appreciate to do it in the GUI than by editing the request URL.

So, I agree with the two additions you're asking for, but I want to 
stress the fact that people who want to use XHTML and to support IE will 
still have to do some dirty hacks.

Regards,

-- 
Etienne Miret
Ne m'envoyez pas de fichier Word SVP, je ne peux pas les lire !
Don't send me Word attachments please, I can't read them!
http://perso.ens-lyon.fr/etienne.miret/Netiquette/no_MS_Office
Received on Monday, 21 April 2008 15:32:49 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 25 April 2012 12:14:29 GMT