Constricting

One switches slowly to the Strict variants of HTML (preferably and for all new documents and sites) or XHTML (if I have to preserve something from the past). One simply wishes to be more correct. It is an ongoing quest.

Back dans la journée, the Total Validator told me I was guilty of using incorrect id values in this blog (short for “Web log”), as apparently a colon is not permitted in such values. Well, that’s news to me. Apparently I should never use lang= and only ever use xml:lang=, which could very well be the stupidest rule in Web development ever. (They both contain the same letters in the same sequence. I’m all for gilded lilies when the gilding actually nets you something, like a prettier sitting room.) Maybe I’ll worry about those later. Yes, I know the foregoing is bollocks in the first place and no longer true based on current Total Validator claims (inconvenient to bookmark, so I shan’t).

In the meantime, I’m trying to get rid of align="this" and width="that", which is not simple.

• Widths: For horizontal rules, which absolutely are structural elements (they delineate a before and an after), I have a set of styles named w plus a percentage, like .w66 { width: 66%; }. I have a great deal of trouble getting this to work with table columns.
• Table captions: There are obscure CSS properties for those. You have to use .captiontop { caption-side: top; } and similar.
• map: My accessibility pages, which require improvement, use an imagemap in the header graphic. Strangely, Strict documents require a map to be inside a block-level element, so I create a superfluous div. Incidentally, text needs to be in a paragraph or something similar if it also appears in a table cell or blockquote (but not li/dt/dd, oddly).
• Image borders: Can somebody tell my why those are on by default anywhere, at any time? Why don’t future built-in stylesheets in browsers turn them off by default? Nonetheless, I had to do so globally. It doesn’t work all the time.

The problem, really, is this blog, which I set up as XHTML Transitional (WordPress is all but impossible to use in HTML mode). I’m over a thousand posts into Transitional at this point, which, while not exactly equivalent to going under the knife for gender reassignment surgery, is at least almost as hard to undo. How does one batch-edit the contents of a database to remove such items as align="left" and the seriously-hard-to-locate colons inside id values? Should I even bother?