If things get seriously broken, I’ll be switching back and forth between the new theme and the old.

First the good news – I have located an honest-to-goodness (if “highly experimental”) HTML5 validator, and after whacking a few niggly bits, we are now officially valid HTML 5 code, at least as far as the front page goes. And we so fresh, we so raw, there isn’t even a badge for it yet!

Every other page, at least every page with images on it, is certain to fail validation at this point. I will seek out the database search-and-replace plugin that I remember downloading, long ago, and fix the post images with that. Back when I got it, it looked far too scary to actually use, so I never did.

And now for the bad news: It has been brought to my attention that this page now looks completely monged when viewed with Internet Explorer (any version), although it must be said, not quite as monged as it did earlier this afternoon. It seems that IE does not support any of the new page elements, like <nav>, <section>, <header> or <footer>, and at least one source I’ve checked suggests that Microsoft won’t be getting with the program anytime soon. No surprise there.

I’ve had a look at the traffic on this site by browser share, and while the decline of Explorer market share is immensely gratifying, it still hasn’t fallen much below 40 percent. So, I have some decisions to make.

First off, I will almost certainly not be downgrading back to XHTML 1.0 to wait for Microsoft to catch up with the civilised world. Every other browser supports HTML5 to some extent, and I’ve yet to find one that can’t cope with the implementation here.

So, most likely this page will load with a splash panel which will offer a screenshot of how the page looks to the rest of us, a choice of links to download proper browsers, and possibly a bit of Microsoft mockery.

Such is the state of my current thinking. If the above strikes any knowledgable person as a particularly bad idea, please make some noise in the comments.

I ran into this before, a couple of years ago, and the options seemed to be:

• Use a less subtle color scheme, which is within the sensitivities of PC screen renderings – which would make the whole page look a bit nasty on the Mac, or
• Serve different images and style sheet colors according to browser platform, and slowly go batshit insane, trying to cover the bases, or
• Leave it alone. This probably isn’t the only site on the net that isn’t seen to best advantage on a PC.

That’ll be why I left it alone, then.

Right, so I’ve rewritten the theme files for the blog, migrating to HTML 5 conventions. The result, by the way, is much leaner code: the total weight of the new version is down from roughly 90KB to roughly 60KB (that’s the total size of all theme files, including such things as the screenshot and some pretty seldom visited page alternatives. The page overhead is much lower still. Quite incredible the number of nested divs and florid class attributes that I managed to throw away — and I haven’t yet done a ghost-class hunt, which I expect will thin out the code even yet still more. The ghost class hunt sounds much more fun than it actually is.

I think I will take tea before getting into that. Do make some noise if you find something on this page not working as perhaps it ought.

Sheila Lennon had the same problem, and set about the irksome task of downgrading Firefox, before discovering a beta version of a FF3.5-compatible version of the add-on, which you can find here.

If you’re a Firefox user, and you don’t already enjoy the benefits of TMP, you really should. You can find the feature list (for the older version) here…

As it happens, I wound up having to create two accounts at Gravatar, one for each of two of my mail addresses. I expect it’s already on the todo list over there, but I definitely think it should be possible to associate an account with more than one email address.

I gather that a new WP2.3 version of the WP-OpenID+ plug-in is also nearly ready for prime time. It’s going to be interesting to see how these various identity management tools work together.

The plug-in works, and overall, it seems like a good, solid beginning.

As far as I can make out, it’s testing against a very broad, “are there any tags in common at all?” sort of filter, and then giving me the most recent posts in the list. In cases where random search engine traffic lights on an old post, the links displayed will potentially lead visitors into more recent posts. This is good, as far as it goes.

The broadness of the positives, however, in combination with the ranking by descending ID number (if that’s what it is) and a set maximum number of links, gives a very constricted set of results.

This is partly due to the way my tags have grown. When I first migrated into WordPress, I used the Categories as if they were Tags, and created over 1,000 of them. Even with a collapsing Category tree, this was unsatisfactory, so in comes Ultimate Tag Warrior (the mightiest WordPress plug-in I’ve ever used, and please, Christine — can has a 2.3 version?). All of the old Categories became Tags. I also added the Categories of each new post as Tags automatically.

Long story short: many of my tags are also Categories, and most of the vintage (pre-WordPress) posts are only tagged with the old Category names I first assigned to them. If I had a spare life in which to do it, I’d go back through all the posts and review the tags, but I don’t. So, I got to wondering how the Related Posts plug-in could be made more cleverer.

Primarily, the plug-in needs to sort the related candidates more intelligently. How about ranking links based on how many tags they have in common with the current post? I would like to see links to any post that has more than, say, two tags in common with the subject, and then however many, up to a maximum total, which have two or fewer tags in common.

It should also be possible to have the plug-in disregard (or maybe score lower) any tags which are also Categories. This would shift the focus onto less generic tags, and a greater likelihood of close relevance.

A-and wouldn’t it be kind of cool if this soon-to-become-sentient-at-this-rate plug-in were to notice when a number of candidate links were tagged with a term that is also a Category, and offer a link to the Category-filtered page, rather than simply list the most recent posts in that Category?

If one was to get all AJAXy, one might also imagine a candidate check-box list of posts in the post edit screen (or maybe in the preview page?) where one might specifically include or exclude links from a suitably ranked list of recommendations?

I have to hold my hand up to being less than a rank amateur at PHP, but it seems to me that something like this shouldn’t present any too many serious problems. Don’t make me learn to PHProperly and write the thing myself!

As always, comments are most welcome — especially from anyone who can explain convincingly why this isn’t going to work, as it might save a lot of futile botheration.