Next message: Terje Bless: "Re: 'No route to host' problem"
Date: Fri, 27 Apr 2001 07:35:12 +0200
From: Terje Bless <link@tss.no>
To: Thanasis Kinias <tkinias@asu.edu>
Cc: Richard Bingham <R.Bingham@ed.ac.uk>, www-validator@w3.org
Message-ID: <20010427080839-b01010701-ee86a285@192.168.1.6>
Subject: Re: Bug: non-working page validates successfully
On 26.04.01 at 11:13, Thanasis Kinias <tkinias@asu.edu> wrote:
>Terje wrote:
>
>>That said, optional checking of URIs, as a separate feature, is
>>something I think would be a usefull addition to the Validator and which
>>might show up at some point.
>
>It seems to me that this is a job for checklink, not the validator.
I dunno really. I'm sort of amibivalent on the issue of /how/ pure SGML
validation the Validator should do. On the one hand we need to make sure it
doesn't devolve into a mere lint, but OTOH it would be nice to add a few
convenience features.
Specifically, URIs have a fairly well-defined syntax and so checking them
should be ok provided it isn't misrepresented as a part of the SGML
Validation pass. Checking if they resolve (as the linkchecker does) is also
more or less within the scope.
Then there is the new world of XML Schemas where we actually have the means
to apply syntactic restrictions on attribute values at a somewhat finer
granularity then the "it's #CDATA" level.
>Much to my surprise, checklink silently corrects the hrefs by escaping the
>spaces (%20) and reports no error. Shouldn't checklink report the link as
>broken, since it isn't a valid URL?
Yeah, it should. I suspect Hugo is on the road for WWW10 ("Paging Mr.Haas.
Paging Mr.Haas. Mr.Haas to the white courtesy phone please.") so we'll have
to wait until he gets back for a fix/clarification.