Invalid characters in network links?

Every night, I run an automated validation against most of the pages of my sites, submiting them to the w3c validator. (The tool that does that is available free on line at the SEO site in my sig block.)

For some time now, the tool was reporting several non-validating pages for each site that, when clicked off the list produced, were 100% valid XHTML. I could manually force a re-run of the validator tool, and would get back results with different pages listed each time.

I suspected some defect in the auto-submission tool, and was going crazy, till I discovered the sad truth that the failing elements were text in the co-op links (which, because they rotate with every viewing, were producing the irregular valid/invalid patterns).

The w3c Validator reports these as non-SGML characters. (The pages in question are ISO-8859-1, but I have some others--which are not run through the validator--that are UTF-8).

Can the network filter non-SGML characters before sending? Or is there an option I have overlooked to select the proper coding? Or do we just need to monitor and detect at out end?

(To me, 100% valid XHTML is important: I strive for it, and put links on my pages by which visitors can themselves check the validation.)

 

You could try str_replace() (PHP, I dunno the equiv in other languages, but it shouldn't be hard to find) on the characters to their XHTML equivilant escape sequence. Or, better yet, Shawn could do that BEFORE serving the ads, that way ads are served with the &escap; sequences.

 

I tried to determine from a Google if that is exactly so, and in the first 20 hits found two opposed opinions.

If you are quite sure that UTF-8 is 100% backwards-compatible with ISO-8859-1, I am willing to take your word and convert a zillion pages from the one to t'other. (I have only had to deal with this with pages that were explicitly in other tongues than English.)

Are you, then, sure?

 

OK, thanks. I'll see if making the switch solves the problem.

It does, though, raise the obvious question: why would anyone ever mark any page as coded ISO-8859-1 if UTF-8 completely subsumes it?

 

I freely admit that character coding has kept me up later at night than anyone aspiring to an IQ bigger than their hat size ought to admit. But I have just never been 100% sure that UTF-8 was 100% backwards compatible. The folk who write such stuff up on the web had best not any aspire to earn a living as a tech writer.