Since starting this worklog, I’ve been experimenting using the screen that WordPress provides for writing blog posts.

It’s good to see every now and then how the other half lives.

And I’m impressed: the blog editing screen WordPress gives you is significantly better than a poke in the eye with a sharp stick, which distinguishes it from some other systems I have experienced.

But, well, not so nice that anyone could plausibly expect me to use it instead of Emacs. So today I decided it was time to get a bit more realistic about how to draft posts, particularly those that are tricky and that I want to get right. That means, I need a way to edit in Xemacs, using psgml.el’s XML mode, and conveniently upload to WordPress.

The editing-in-Xemacs part is easy. All I have to do is stop not doing it. What was needed was a stylesheet to translate from the version of TEI Lite I habitually use for writing, into the odd mix of normal XHTML and blank lines for paragraph breaks that WordPress uses. (I should probably experiment with using normal p elements — will WordPress let me use them?)

Answer: yes. So I don’t have to render paragraphs using just blank lines. And I didn’t really need to do a new stylesheet; I could just have used my existing TEI-to-HTML translation. I’ll comfort myself with the thought that the new stylesheet is smaller and has less cruft.

It’s still a little awkward that I have to cut and paste from my Emacs buffer into the WordPress write-post screen; I notice that David Carlisle seems to be able to edit existing posts from within emacs, using T. V. Raman’s g-client package. This makes me jealous, but not jealous enough to drop everything to work on an emacs mode for posts to WordPress. (Everything I haven’t already dropped in order to play with my tei-to-wordpress stylesheet and to write this post, that is.)

When the cut-and-paste gets annoying enough, I’ll take the time to make the stylesheet generate the post in Web-forms submission format, and write a script to upload to the server. But for now, being back in emacs for editing is enough.

Installed Spam Karma 2, today, and hope it deals well with comments from real readers (who are welcome here) and from spambots and spammers (who are not).

There are a few points that have caused a little concern.

When I ran the filters on the accumulated comments and spam, it took half of the spam comments and approved them as OK.

I’m going to give it the benefit of the doubt for now, assuming that it’s using some sort of Bayesian filtering that needs training. If this is still happening a month from now, it will be harder to be patient.

You can of course review approved comments and say “No! bad choice! that one is spam.” And similarly you can review the spam bucket and say that something is not actually spam.

It was at this point that I almost turned it off again, after ten minutes of use. It’s set up to display the lists of recently harvested spam and approved comments in a concise tabular form, with the admirable goal of getting more data onto the screen, and to expand any cell in the table if you hover over it, so as to show the full entry.

Perhaps my browser is set up to call the hover() method more quickly that the developer expects, but I found it impossible to move my mouse across the table (e.g. to reach the scroll bar, or to reach a cell I wanted to hover over) without one cell after the other exploding, shifting text around on my screen, and then collapsing again. It reminded me of nothing so much as flying as a child in a Cessna or Piper Cub and hitting air pockets that caused the plane to fall precipitously, climb again, and then drop again. I can almost smell that unforgettable stench of high-test fuel in an small enclosed space.

The only way I could reach something I wanted to hover over and look at was to sneak up on it by moving my mouse cautiously down the left or right margin and then into the table. And even then, it only expands one cell at a time, so you don’t get a good overview of any individual comment and its content and karma rating.

Fortunately, you can turn the hover behavior off. You have to turn on ‘Advanced’ options and uncheck the box that says “if using … css-impaired browsers, you should disable this feature”. It doesn’t mention susceptibility to motion-sickness.

The table of recently harvested spam is labeled as including comments with karma lower than -20; the threshold can be changed, and I’m glad to have a clear statement of what’s displayed and what’s not. But I found it confusing to have comments with karma of -6 and whatnot included. Eventually I concluded that the screen meant “higher” than -20.

The beauty of open source is that I was able to correct that typo without any significant effort.

The display of information about recently harvested spam with bad karma is as close to illegible as a mere color scheme can make it. (Well, that’s too harsh; I’ve seen worse, but not in a color scheme intended for actual use.) Dark green and medium-dark red against dark gray.

The beauty of open source is, again, I can change it. I’ve got CSS and I’m not afraid to use it.

And finally, the author has announced that he doesn’t expect to update it, because he’s tired of incompatible changes to WordPress that break plugins.

I can see his point. But any problem caused by his decision is down the road a while; perhaps others will pick up the code, or perhaps I’ll shift to a different tool later, just for educational purposes (the way I change Emacs color themes every few months or so, to see how they wear over time).

On the other hand, despite the concerns, Spam Karma appears quite popular and successful, and to have attracted fewer vehement denunciations than some other anti-spam tools. So the official attitude here is the traditional guarded optimism.

We will learn, as time goes by, whether that optimism is justified.

[15 January 2008]

This morning I was a bit surprised to see nineteen comments on various blog posts waiting for moderation here. I hadn’t realized that such a queue had been built up since Friday, or that so many new people had discovered this blog and wanted to comment.

My slight flush of self-congratulation at the success of the blog lasted only a few moments, as long as it took to notice that of the nineteen comments eighteen were clearly spam, and the nineteenth is almost certainly spam (I’m going to sleep on that one).

Dealing with persistent spam attacks could take a lot of the pleasure out of maintaining a blog, just as dealing with spam made email an unbearable drain on my time before I started my white list. Time to spend some time investigating the blog-spam control tools available to me.