Re: IO error

"Michael[tm] Smith" <[email protected]> Fri, 26 Jun 2015 12:43:54 +0900

Newsgroups	gmane.org.w3c.validator
Message-ID	<[email protected]>

Alberto Nannini <[email protected]>, 2015-06-19 00:43 +0200:
> Archived-At: <http://www.w3.org/mid/[email protected]>
> 
>    your validator rise up an "IO error" validating my site.
> 
>    Here is my homepage url: [1]http://www.notanumber.it/en/
> 
>    Let me know if there is something wrong.

That server is sending a broken gzip response.

I think the response is either missing the standard/expected 8-byte gzip
trailer, or the gzip trailer is corrupted—maybe because the server is
closing the connection before it’s sent the whole response.

If you have the “curl” command around, you can test the server by doing this:

  curl -s -H “Accept-Encoding: gzip”  http://www.notanumber.it/en/ | gunzip -t

You’ll get back the error message “gzip: stdin: unexpected end of file”.

You won’t see that error message if you try the same thing for a URL from a
server that’s sending a valid gzip response; for example:

  curl -s -H “Accept-Encoding: gzip” https://sideshowbarker.net/ | gunzip -t

The reason this doesn’t cause any problems in Web browsers is that I think
browsers will quietly ignore a corrupted or missing trailer. But other
tools (e.g., the gunzip tool) will not. So the right solution to the
problem would be to try to get that server fixed so that it’s not sending
out a broken gzip response.

In the mean time if you want to run the HTML checker on any URLs from that
server, you’ll need to turn off gzip compression on the server (until you
can get the cause figured out and fixed).

  —Mike

-- 
Michael[tm] Smith https://people.w3.org/mike

signature.asc (application/pgp-signature, 819 B)