Older blog entries for markonen (starting at number 32)

dyork: The height of a small cap should be equal to the x-height of the typeface. In most cases, the difference between the x-height and the height of an uppercase letter is big enough to make “fake” small caps look bad even to the casual observer. The obvious problem is the difference between stroke widths of the cap and the small cap, but there are more subtle differences as well. Most real small caps cuts have tweaked the “optical size” of the letterforms to suit the smaller size of the characters.

Unlike haruspex, I don’t really think that this is your problem. It seems to me that the correct browser implementation would be to render the small caps only if there’s an appropriate small caps font available, and revert to normal rendering otherwise. Apparently nobody is doing it this way, though…

Okay, credit where credit’s due: in the previous post I talked about Apple’s default installations of cyrus and PHP being substandard. Their bundled python and perl installations leave little to be desired, however. It’s pretty amazing to see PostgreSQL’s PL/Tcl, PL/Python and PL/Perl building and running on an out-of-the-box consumer OS without any additional software.

I do have one wish for the next Mac OS X release, though: please, please, please bundle readline. Much of the already included software would find and use it, resulting in a much better cli experience.

I’ve managed to infect my IMAP account with a hierarchy of folders that cyrus isn’t letting me delete on-wire. Normally this wouldn’t be a problem, but alas, I’m on Mac OS X Server, so it is. Apple has built a semi-usable GUI tool for mail administration, but you really can’t do any low-level operations with it. Nevertheless they’ve decided that it’s enough, so they ship the whole thing without a working cyradm. What’s up with that?

I love Apple’s practice of using proven open source tools as a basis for their products, but crippling the results like this isn’t making anyone happy. (The same is true of Apple's bundled version of PHP, which is configured in a particularly useless fashion.)

Back in 2002 I advocated a new kind of a banner ad that was cheaper to make, got better results, and didn’t annoy people enough to create a market for banner blocking software. The idea was to go from small banners, strobing at 20Hz, to bigger but static images. After all, the approach seemed to work pretty well in magazines and newspapers.

Well, I’m happy to report that the future has arrived — DoubleClick is selling space for 160x800 pixel “Giant Scyscrapers” that, as a plain old JPEGs, work really, really well. Here’s a real Neiman Marcus ad for Marc Jacobs, as displayed at nytimes.com. I can’t wait to see other advertisers join the bandwagon.

I’ve spent this week working on a Wiki-based content management solution for a customer of ours. While browsing through one Wiki after another, the most baffling thing on the scene is the total domination CamelCase has over Wikipedia’s much more readable syntax. And frankly, it’s not the only thing that makes me wonder whether the value of good writing even registers on most wikiheads’ radars. In edit wars, content structure and language coherence are always the first casualties.

In other news, I’ve hacked my real blog’s RSS feed to mirror the advogato one. I haven’t updated the “real” site in a year, though, so if someone is still subscribed… wow!  Be prepared to see my ramblings a bit more often in the future :-)

2 Feb 2004 (updated 2 Feb 2004 at 18:50 UTC) »

I’m sure Microsoft was happy to pay for the bandwidth and disk space this little gem takes on their developer site: “Small companies should stay out of markets that are big enough to be interesting to big competitors. … You can’t beat big companies. The best way to win a fight is to not be there.”

23 Nov 2003 (updated 23 Nov 2003 at 23:13 UTC) »

I just read Aaron Swartz's piece on procrastination. The subject has been in my mind a lot lately. I've always been a procrastinator, and it's been a real problem for years and years.

Aaron did a great job in luring me to think about this in a structured way, and I already came up with a great example: RSS readers. Intuitively, you'd think that an RSS reader would increase your productivity by decreasing the amount of time spent checking if a blog has been updated.

But I've noticed that's not really true. My RSS reader, NetNewsWire, has instead transformed catching up on blogs into a distinct task: getting the unread count back to zero. I've now realized that I'll always choose that task over real work. And it's the same with email. The only time I'm getting work done is when those unread counts are gone from NetNewsWire's and Mail's Dock icons.

I'm now running both my Mail and NetNewsWire primarily on my laptop, so my main development system should be free from such attractions.

Now, back to work.

haruspex: That's not really true; Office only costs about four times as much as Keynote, which I think is pretty fair since Office includes four pieces of software of comparable complexity (Word, Excel, PowerPoint, Entourage). But your point is generally valid. The license is just an additional data point to use when comparing software.

Also, to bring this more on topic, it is clear to me that license management itself is so resource consuming that the promise of getting rid of it altogether is perhaps as great a selling point for free software as the zero cost for licenses. But this approach also places an emphasis on the level in which a particular free software license is "hassle-free", and I think that the various BSD licenses will fare better than the GPL in the long run because of this.

18 Nov 2003 (updated 18 Nov 2003 at 01:22 UTC) »

I got a cute new twelve-inch PowerBook last Friday, and promptly proceeded to install all my software on it. It turns out that there is some variation in commercial software licensing here: some licenses are per user, not per machine, others explicitly allow an additional copy to be installed on a laptop, and some require you to go out and buy another licensed copy. Apple's turned out to be the only commercial software I have in the latter camp; they need separate licenses for the PowerBook. This can effectively double the price of a product when compared to its competitors, so it's worth looking into when shopping around. Here's a rundown of the commercial software I looked at, and their licensing:

As you can see, Microsoft's aren't the worst software licenses out there! Apple clearly beats them here in customer hostility. Another thing you can see is that I need quite a bunch of commercial software in my work. rms says that we should be willing to sacrifice convenience and functionality for freedom, but in reality you'd also be sacrificing productivity. And that, frankly, can be too high a price to pay.

14 Nov 2003 (updated 14 Nov 2003 at 23:53 UTC) »

My home network connection has been really sluggish as of late, and I was actually going to call my ISP about it when I realized the real culprit: WPA.

I don't really need encryption on my wireless network since basically everything I do is encrypted on the application level. I do want to keep a level of privacy, however -- our mail server, for example, is an effective open relay when accessed through this network. So, I've used WEP until now, but the 128 bit shared keys are really, really cumbersome to enter manually in the hexadesimal format. So there was much rejoicing when I was able to upgrade to WPA last week because Apple added support for it in Airport v3.2 software. My Linksys WRT54G base station already supported it, so I was good to go. Say hello to human-readable passwords.

Here's a ping report to the Linksys from my Mac 10 feet away:
209 packets transmitted, 204 packets received, 2% packet loss
round-trip min/avg/max = 1.391/14.665/57.615 ms

Needless to say, it's back to WEP for me as soon as I can figure how Apple mangles a human-readable password into a 128 bit shared key. I could then enter the said key to my Linksys and continue to use the password on my Macs. With a bit better performance.

Update: The Linksys just needed restarting to regain normal performance. I can't help thinking that this is the downside of basing embedded devices on desktop operating systems (Linux).

23 older entries...

New Advogato Features

New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

Keep up with the latest Advogato features by reading the Advogato status blog.

If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!