Re: [Htmlvalidator-help] [HTML Validator] Validating HTML behind a proxy

[Marc G. <mg...@sk...>]

Hi Andy,

Quite interesting mail.
You are in short quite right. The SGML parser is builded inside a DLL 
called nsTidy.dll (or .so)
containing both tidy and opensp.

It bundles too most DTDs of W3c to make validation locally. But I think 
that opensp has the
bad habit to make anyway request to a DTD when the link is used in the 
doctype. I fixed a
lot of issues about this. Because I work also most of the time behind a 
proxy.

I do not know the issue you speak about here. I would be really 
interested to get a sample
to see what I can do to solve it.

Thanks by advance,

Marc

Andy Dingley wrote:
> I love the extension, especially the new SGML validator. However we've
> also just switched to a web proxy here in the office which kills it
> dead!
>
> I've read a past message, suggesting that the problem is with OpenSP not
> supporting proxies for DTD retrieval and that there's little you can do
> about it. This is sad, but understandable.
> http://sourceforge.net/mailarchive/message.php?msg_id=37500793
>
>
> I've noticed something though, which gives me an idea. It seems possible
> to validate HTML pages with a doctype declaration that only use the
> identifier and not the URL. I assume that your validator has some
> "well-known" DTDs (such as HTML 4.01) embedded into it, and for these it
> passes them directly to OpenSP, rather than requiring them to be loaded.
>
> Would it be possible to also do this for the well-known DTDs, even if
> these also specify a URL ?  Obviously this might be restricted to those
> that also specified an equally well-known URL known to match that
> identifer.  Even if it's no help to people with those funny
> editor-specific HTML 3.2 extended doctypes, it could be a great boon to
> those of us trying to hammer out mainstream HTML 4.01
>
>
> As it is, I've recently been trying to sell the local pointy-haired boss
> on the virtues of "valid HTML", only to have him then reject my recent
> page as having "1484 HTML mistakes in it" (mostly "element <p>
> undefined" - it is of course perfectly valid). Ah, preserve us from the
> User with Half A Clue...   8-)
>
>
> Thanks again for a great piece of kit
>
>