Re: Reporting a Validator bug

John B. Robb wrote:

> Thanks for your detailed response.  I take it then, that there is in
> fact a bug in Validator.

Well, virtually any nontrivial (i.e., useful) software has bugs.

Did you miss Ville Skyttä's message? He replied:

"Thank you for the report.  This appears to be a bug in the XML-LibXML 
module
used by the validator which causes it to end up in an infinite loop:
https://rt.cpan.org/Public/Bug/Display.html?id=56671 "

So the problem has been identified and they are working on it.

Meanwhile, you might consider using the validator
http://www.htmlhelp.com/validator/
which has no such problem. It reports that your page is valid but informs:

"Line 63, character 74:

...  of William of New Kent &#133; Albemarle Cos, below]</p>
                                 ^

Warning: reference to non-SGML character"

Technically, this is a warning only, since &#133; is not invalid, just 
undefined. Practically all browsers, or almost all browsers, interpret it as 
meaning the HORIZONTAL ELLIPSIS character. But you can avoid the issue by 
using the entity reference &hellip; instead, just as you use entity 
references like &ldquo;.

> Your suggestion to use images to simplify the code might make it look
> neater but adding hundreds of images, one for each line, would make
> loading impossibly slow.

It was a questionable suggestion and rather off-topic. This validator list 
is not for discussing page design issues.

> I have left the line
>
>   <a>Last updated 14May2009<br />&copy; John Barrett Robb</a>
>
> as it is, even though it has no "href", "id", or "name" components,
> because I don't need them.

This seems to be related to the following text:

> Errors I found
[...]
> LINE 663 An "href", "id", or "name" attribute should normally be used
> with this "a" tag. If one of these attributes is
> being set with a script, then consider the accessibility issues of
> requiring scripting. Disable this message if you do
> not want to check for this.

This, and other similar texts, were apparently the output of some 
unspecified loose checker, or "lint-like" program. It's very confusing to 
present such stuff in this list even without telling what is being quoted.

Formally, there is nothing wrong with using <a> without attributes. It's a 
bit debatable what the semantics is then, though. And such elements are 
somewhat risky, since some style sheets or program code might be based on 
the (questionable) assumption that every <a ...> element is either a link or 
a destination anchor. There's also the risk that someone reading your HTML 
code will think it's an error - and maybe generates some real errors when 
trying to "fix" it.

I don't see why you are using <a>, partly because I cannot access your style 
sheet. As such, <a>...</a> markup has no impact on anything, so why not just 
omit it? If you need an inline container for some reason, why not use just 
<span>...</span>.

At the end, you quote your original message:

>> I have a webpage (attached) with an extremely simple structure which
>> the W3C Validator chokes on. [...] It is unusual only in that it
> contains a large number of "&nbsp;"s, which I need to control the
> exact spacing of an indented genealogical tree chart
> (presented like an indented outline).

Two remarks:
1) You should post a URL, not an attachment. In particular, the URL lets us 
access the HTTP headers as well as things referenced by relative URLs.
2) &nbsp; does not guarantee any specific exact spacing. CSS has much better 
tools for fine-tuning spacing. Of course &nbsp; is valid, irrespective of 
the motives for using it, but validity is just a formal aspect.

-- 
Yucca, http://www.cs.tut.fi/~jkorpela/

Received on Saturday, 17 April 2010 19:34:09 UTC