Back when I was using Blogger, I used to have a linkblog called, cleverly, “Geekfun Links.” I tried to use Geekfun Links as a spot to blog stuff I thought was cool, but wasn’t going to say much about.

I’ve decided to dust it off again and wire it up to the cool blogging integration in Onfolio’s new version 2 beta. I just wish their blockquoting of text snippets was better.

So, one of the problems with stuffing your mouth with pistachios as fast as you can shell them is that you might still have the meat of several perfectly good but under-chewed and unswallowed nuts in your mouth when you start to chew a bad one. What’s worse, it might take a couple of chews before the bad one is really obvious, by which time you’ve probably added a few more nuts. The result is that you have to spit it all out, and then it might take a few more nuts before the flavor is really purged.

This is not unlike the branch-prediction miss penalty suffered by the Pentium IV processor. Only rather than chewing on nuts, its chewing on instructions and data.

I’d like to set up a Wiki, but which one?

I installed MoinMoin about a year ago but didn’t end up doing the project I installed it for and at this point would rather try something new.

Basically, I’m looking for;

1) Easy to install
2) Attractive URLs. MoinMoin’s URLs looked like crap in the CGIWrap environment I ran it in.
3) Ability to attach files
4) Ability to include images
5) Ability to set read/write permissions on subtrees.
6) Versioning without a dependancy on CVS
7) RSS Feeds of changes
8) Inclusion or plug-in integration of a simple WYSIWYG editor.
9) Abilty to do Wiki level read/write permissions
10) Python, PHP, Perl, but not Java due to the constraints of my web-host.

Also nice but not important.
-Sortable/filterably lists
-Structured Data in pages
-Easily created aggregate view of pages with Structured data.

Any suggestions?

I got sick of comment spam a few months back, so I did a half-assed job of dealing with it. I installed mt-blacklist and disabled comments on new posts, but did little or nothing about closing comments on older posts.

Since then I’ve had to spend 15-20 minutes a week deleting comment spam that slipped by mt-blacklist and got added to old entries that still had open comments.

I’d made a few false starts at turning off comments on old posts en masse in order to reduce exposure but kept stumbling over the fact that most of the solutions seemed to be for MovableType 3.x and I am still on 2.66. I finally fixed that though, by installing mt-close, which lets me close comments on all posts older than a certain number of days, which I’ve now done.

The result is that I seem to be catching a lot less comment spam and so far the only new comment has been a legitimate one. On the flip-side, I’ve seen my first bogus trackbacks, but so far, only a few.

The good news is that now I have the tools in place to make managing comment and trakcback spam managable, so I’m opening new posts for comments again and I’ll only close them when the rate of comment spam gets high enough that it takes me more than a minute or two a week to deal with it.

I’m now state-of-the-art for 2003! (if one ignores my barely customized template).

Back when the beta arrived, I wrote that Google Desktop Search is a disappointment because it didn’t support indexing Firefox history or Thunderbird mail. Later I praised its good points.

Now that the product has been released, I thought I should follow up. First off, the seem to have retained the good points. It remains a tiny download, even with added features; it is fast; and it doesn’t hog a lot of resources when it is idle, at least on my home desktop.

They also addressed my main gripe by adding support for indexing Firefox History and Thunderbird e-mail. And while they didn’t address my desire to index my blog, they did provide a Plugin SDK, which has allowed someone else to create a plugin called Kongulo that can index web sites which should meet that need.

One notable feature of their plugin SDK is that it can handle compound files and index each item as a separate record, for example, indexing each row in a spreadsheet or database separately. The desktop search products from Microsoft and Copernic also have APIs, but appearantly have to index the entire contents of a file as a single record.

Of course, now that I’ve got the full version, I note that it has other shortcomings. For one thing, its still a bit jarring to see items from my PC listed in the same page as a Google internet search. That I can get over (or turn off).

Harder to deal with is that GDS happily indexes the spam folders in my e-mail. I’d like to be to exclude mail folders easily, but according to their support, its currently not possible.

I’ve also noticed that when I tell GDS to open an e-mail from my search results in Thunderbird, it opens the wrong message.

The fact that the search interface is ordinary HTML also has some frustrations. It makes it difficult impossible to quickly resort and scan search results by other meta data like sender, recipient or date.

In any case, I’m hapy to see GDS supporting the browser and e-mail I use, and I look forward to Onfolio releasing a plugin so I can search my collection of web clippings at the same time I search e-mail and the like.

One Firefox extension I have no intention of uninstalling is SessionSaver. SessionSaver keeps track of the state of your browser — window positions, open URLs, scrolling positions, even tab-histories — and will restore them whenever you restart the browser, even if Firefox crashed. Next up is to install it on my wife’s computer — she keeps a lot of Firefox windows and tabs open, which really sucks for her every week or two when Firefox or an errant plugin finally freezes or blows up.