- From: Frank Ellermann <nobody@xyzzy.claranet.de>
- Date: Fri, 07 May 2004 07:12:35 +0200
- To: www-validator@w3.org
- Cc: olivier Thereaux <ot@w3.org>
olivier Thereaux wrote: > we may be able to expand our list of supported charsets. Good. I'm not planning to publish PC-multilingual-850+euro or similar pages, but IIRC somebody here wanted codepage 437, and somebody else MacRoman. [bugzilla] > What trouble can you think of? It's on my short list starting with "sed", and that's something I'd never ever show normal human beings... ;-) In fact I was even unable to change my old address on your bugzilla version. > I thought gmane was mostly a read-only thing, was I wrong? It's a Gateway MAil NEws, and apparently write access on this list (W3C validator) works. Web archives are fine for online reading, but difficult to use as soon as I want to write to a list, [source listing for errors] > Any advocate of keeping the current behaviour? As an option it's fine, but _only_ as an option. Maybe using a cookie with preferences. I'd then set my preferences to NO ECHO because most of the time I try to validate my own pages. But sometimes I try to find out why something doesn't work with my browser, then it would be nice to get the (foreign) source with the report. >> "DOCTYPE override" info _can_ appear twice in the output [...] > Do you have a test case for that? <http://purl.net/xyzzy/home/test/res.htm> with HTML 2 results in two "DOCTYPE override" warnings. Maybe I shouldn't force "HTML 2" for "valid HTML 2 strict" - this was only a test ;-) [HTML 2 icon] > Need to look into that Same URL without override => no icon. Almost irrelevant from my point of view, but it could be some kind of general DOCTYPE confusion in the validator causing both "errors". ["start your httpd" error message] > test case for the (strange) message you're quoting? Try an URL pointing to an existing host without a http-server, e,g. abuse <http://testcase.boulder.ibm.com/> and validate it with option "test error pages". > without knowing the encoding, finding the doctype is a matter > of garbage-in garbage-out, and it's therefore not logical to > adapt the mechanism in case of absence of charset to an > hypothetical detection of the doctype. Yes, that makes sense. OTOH I don't like any overhead in error pages. There should be a way to validate very simple documents like "<title> oops </title><pre> go away </pre>", where DOCTYPE and charset are irrelevant, as long as the text doesn't contain unintentional "<" or "&" characters or similar stupid errors. For "irrelevant" read "the oldest DTD you have and Latin-1" ;-) [fragments] > Indeed, fixed (in CVS). Tnx. How did you do this, there is now a name="skip", but it's still valid XHTML 1.0 strict ? Or do I confuse 1.0 strict with 1.1 ? BTW, the "results" link still doesn't work for _valid_ documents, because then you have no "results" fragment at all, but that's of course no problem. I haven't tested this effect with your list of "strange cases" in the test suite. >> ",text" and ",tablin" are different documents. > Agreed. Suggested fix? Hard to say, because I see the version without CSS. Is there any reason why you don't offer text / tablin results as options on the extended interface ? Otherwise you could implement this as options without mentioning it on the normal result page. >> - "help and FAQ" >> <http://jiggles.w3.org/servlet/docs/help.html> >> is apparently broken. > Weird, where did you find that link? No idea, I can't reproduce it today, forget it. [Help for options] > This is now done in CVS. Tnx again, but please add a name="Options" on this page. Maybe it's only my browser, but if it goes to a new page and doesn't find the specified fragment, then it displays the _end_ of the page. And scrolling back I would first find "output options" instead of the wanted "options". All in all, I like the new version, the only serious problem is the forced source listig for errors. Bye, Frank
Received on Friday, 7 May 2004 01:20:38 UTC