BibDesk 1.3 was just released (see here for very complete release notes), and this release adds a few features that were requested four years ago - searching PubMed and libraries with Z39.50 support. If you use those resources, this means no more downloading files to import references - just search from within BibDesk, and your results show up as a group that you can browse and through with the same interface you use for your local references.

Other notable features include using URLs or scripts as sources for external groups - if you have a script that can generate data BibDesk can read, its output can show up as a group.

Two new data formats are now supported for importing - MARC (and some MARC XML) and Refer. Dublin Core XML is also supported.

Note that this is also the first version that will no longer launch on Mac OS X 10.3.9.

Kudos to everyone involved - check out the contributors wiki page for who to thank!

I couldn't make it to WWDC last year, I'm not at Macworld this week, but I will be making it up to LA on the 19th for the Leopard Tech Talk, to catch up. I'll probably be there the day before, owing to traffic and an inability to wake up early.

If any area mac devs are meeting up around then, drop me a note.

I wonder if we'll know anything by then about developing for the iPhone.

If you're like me and don't have a Leopard Preview, and if you haven't seen the Leopard Developer Application Technologies Overview, you should take a look. There's some pretty interesting stuff coming down the pipe. I like the Calendar store, which lets any application work with the iCal calendar info in much the same way as AddressBook.framework opens up the Address Book. Maybe we'll see some enterprising developer add support for travel time in calendar display?

I also like the Applescript Bridge - an idea whose time came YEARS ago. It's buried under the section titled "Picking Up the Pace of Cocoa Application Development". What a major simplification!

I like that a lot of the new advances in Leopard seem to involve making it easier for apps to work together and share data. My data should belong to me, not to the application I first entered it in. This is as much a usability issue as it is a data safety, vendor lock-in and openness issue.

A great article about Burnout and "Hurry-sickness" from New York magazine - "Where Work is a Religion, Work Burnout is Its Crisis of Faith" Interesting, but thankfully not as relevant now as it would've been a year ago.

One quote I loved (and would love to see proof of:)

Elevator engineers even have a term for how long it takes—door dwell—before people start jamming their fingers on the door close button, which is usually a placebo, a function already disabled by litigation-conscious building managers

This reminds me of an idea I had in high school to add a "Go Faster" button to Netscape's toolbar. It wouldn't do anything, but it'd give you something to do while waiting for the Cool Site of the Day to load.

Found via Buzz's delicious feed.

The GTD App Midnight Inbox 1.0 was released recently, and it has a number of interesting features, including automatic harvesting of 'inbox' materials like email and desktop files, and reminders to get back to work.

I also really like the graphics and typefaces in the UI - I really wanted to love this program.

I tried it a few times during its beta period and never quite understood what was going on. Buggy interaction kept me from experimenting enough to get it. The 1.0 release squashes most of the UI bugs that I was running into, and I got the feeling that I could work with this program. There were two big problems that stopped me from diving in (I didn't buy it) - no documentation, and speed.

In large part, the usage model explains itself (it helps to have GTD experience). But it doesn't completely explain itself, and the lack of basic documentation leaves you without a guide to the program's subtleties. The small "Introduction" window reads more like advertising than a manual, and that's all you get. The best you can do is explore by double-clicking everything, and typing away.

There's no clear explanation why two kinds of text notes exist, how to deal intelligently with task times, or why you might want to use a sheet to review instead of just going through things yourself.

Can I drag this there? Can that have sub-items? Can I edit that? What does it mean to make a context active? What's that moving grey bar in the top part? Your guess is as good as mine.

I could temporarily forgive the lack of a manual and dig through forums (for example the 'inboxbeeps' google group) and emails to figure these things out - if it weren't for the speed. Maybe progress has moved past my G4 Powerbook and everyone reading this will notice no problems at all, but Inbox was just too slow for me - it had responsiveness failings like delays in creating new actions and notes, delays in recognizing clicks to edit, losing keystrokes while editing a project name, and the killer - a spinning beachball while dragging a project to reorder it.

So, I think it's got potential. It definitely follows the GTD system. I loved the super-easy 'quick note' feature, accessible from anywhere. If it got a lot faster (or I got a new machine), and then - if it got some real documentation, I'd give it a serious try. But for now, I'm going back to Kinkless.

Note: (10 minutes later) I just saw a note in the inboxbeeps group that they're working on docs now, and I've forwarded my list of questions to the authors. Hopefully this will be helpful, and then my only remaining criticism will be the speed.

After years of being aware of David Allen's GTD system through 43 Folders and other sites, I took the dive about a month ago and dumped all my various to-do lists and project ideas into Kinkless GTD. I had tried to keep reminders and actions in a variety of other systems, including emacs org-mode and VoodooPad, which I use extensively for notes.

I never kept up a system for very long, and I believe the contexts were the key problem - having to look at tasks with different contexts in the same list is daunting - I end up not writing things down to avoid a frightening list, and keeping multiple lists around is difficult in every system I tried, until I tried Kinkless. Organizing by context gives a great feeling of confidence when I ignore huge lists that I just can't do right now. Knowing what you can't do makes getting the rest started much easier.

I'm not going to write a big series of posts describing my system or anything - I might have something to say about contexts later, but I would like to say that "vanilla" GTD can work for programmers, graduate students, etc - without major changes. Avoid the temptation to focus on your methods and just try using the simplest thing you can get away with. It'll probably work.

I've updated AddressService to a Universal Binary, and relicensed it under the LGPL.

AddressService is an OS X system service that lets you select a part of a person's first or last name and replace it with their addresses or phone numbers. It's handy for quick entry into an email or chat session.

I haven't been able to thoroughly test it on an Intel mac, so please let me know if you find any issues.

Last spring, when I wrote about highlights from the NHL Playoffs being available on the iTunes store, I said:

the sensation I felt when I noticed the announcement of NHL playoff video available on iTunes was that of Steve Jobs signing my name on a check for ten dollars and winking.

This weekend, when I saw this: video.google.com/nhl.html - free videos of every NHL game, including historic games, I went plain nuts. Go check it out, even if you don't know you like hockey yet. If you don't have a team, may I suggest the Pittsburgh Penguins? There's probably still time to become a Penguins fan without being a bandwagon fan, but it's running out...

It's been mentioned before, for example on Digg, but iTunes 7 has a new feature that counts the number of times you've skipped a track, which is defined as playing the song for at least two seconds but no longer than twenty (after which it is counted as played fully).

I'm curious how they chose twenty seconds as the cutoff - when I was working on an (unreleased and unfinished) project for generating smart shuffles, I defined skips as not listening past the first fifteen seconds, but didn't count a full play until the song ended.

It's an interesting feature, and it's apparently kept up to date by some iPods, so I can see it helping me keep bad songs off my small iPod by using it in a smart playlist rule. However, often I find myself skipping songs I really like because I'm just not feeling them at the moment. So - how about the option of a smart playlist rule that lets me define a skip percentage? If I skip a song far more often than I play it, then it's a pretty good bet I don't need it on my iPod.