Re: EOF detected with HTML5 and NFC

Hi Ville,

> @2010-04-19 22:17 +0300:
> On Monday 19 April 2010, Michael(tm) Smith wrote:
> > Julien ÉLIE <julien@trigofacile.com>, 2010-04-07 18:26 +0200:
> > > I'm checking my web site with the new HTML 5 parser and I do not
> > > understand the following error on:
> > >    http://www.trigofacile.com/jardins/lucullus/grec/democrite/fragments.htm
> > >    "Line 123, Column 246: End of file seen and there were open elements."
> > 
> > I can reproduce that error when I use the HTML5 facet of the
> > http://validator.w3.org/ UI, but I can't reproduce when I validate
> > your document directly against the backend validator.nu instance
> > we have running on that host.
> 
> I tested it a bit.
> 
> With my local validator instance which always posts to validator.nu I cannot 
> reproduce the problem.  On qa-dev.w3.org which posts to its local HTML5 
> validator instance it occurs apparently every time.  But if I reconfigure the 
> qa-dev validator instance to post to validator.nu, the problem goes away.  (I 
> reverted the qa-dev validator instance at http://qa-dev.w3.org/wmvs/HEAD/ to 
> post to the local qa-dev HTML5 validator for testing so it fails again at the 
> moment.)  On the other hand, consistent with your findings, directly using 
> http://qa-dev.w3.org:8888/ does not show failures.
> 
> Could it be that some of the validator.w3.org boxes have a buggy/outdated 
> local instance of the HTML5 validator where the issue occurs only when using 
> the POST body input interface?

I have the validator.w3.org hosts up to date with the same code
that validator.nu is running (that is, the latest upstream sources
from the validator.nu repository).

When I do a post (using curl) to the port 8888 instances from the
shell on those hosts, I cannot reproduce the EOF error; instead,
the document in question seems to processed as expected. The
following is the complete message output I get:

  # curl -s http://www.trigofacile.com/jardins/lucullus/grec/democrite/fragments.htm > fragments.htm \
    && curl -s -F out=gnu -F charset=utf-8 -Fdoc=@fragments.htm http://localhost:8888
  "fragments.htm": info: The Content-Type was “text/html”. Using the HTML parser.
  "fragments.htm": info warning: Overriding document character encoding from none to “utf-8”.
  "fragments.htm": info: Using the schema for HTML5+ARIA (experimental).
  # 

That is, it validates without any error messages at all being emitted.

If you can think of any other troubleshooting I could try there,
let me know.

  --Mike

-- 
Michael(tm) Smith
http://people.w3.org/mike

Received on Sunday, 9 May 2010 12:03:01 UTC