Weblog Without Honor or Humanity

IronPython is a Python implementation targetting CLR (i.e. .NET platform as implemented by Microsoft or mono).

Interesting tidbits:

• it’s 1.7x faster than Python 2.3 on standard benchmark (although supposedly only barely faster on real-life examples)
• the author has been hired by Microsoft to work on CLR team where he’ll continue improving CLR support for dynamic languages with IronPython being a test-bed but production-quality implementation

Things get really interesting!

Linus has spoken: using 0 (instead of NULL) to denote a null pointer is wrong:

char * p = 0;	/* IS WRONG! DAMMIT! */
int i = NULL;	/* THIS IS WRONG TOO! */

I concur.

Korean mobile carrier is planning to stop providing Brew.

Good riddance, Brew is just another closed platform that hopes to make easy money thanks to holding complete lock on services offered through it (therefore a lock on innovation).

The news aren’t that good, though. Brew will just be replaced by another closed platform developed in Korea.

When will they (mobile carriers collectively) that the way to win is by offering open platform and letting other business to freely build and sell services on this platform. Long term this strategy outperfoms any “closed garden” wet fantasies of carrier executives. The problem is that they just can’t stop charging for SMS.

Chicago Suntimes has published a positive article on WikiPedia. It’s good to see mainstream acceptance of the radical concept of encyclopedia produced by what many consider to be an equivalent of a million monkeys.

The author admits that the quality of WikiPedia articles is higher that what he had expected and names WikiPedia “one of the Internet’s top 5 information tools”.

So what’s the obvious conclusion from Microsoft’s Lookout buyout? Mozilla Thunderbird needs a built-in, full-text search too.

Given Thunderbird’s C++ nature, C++ port of Lucene would be the best code base to integrate with my favourite e-mail client.

Review of Hot Text - web writing that works.

Verdict: if you write for web, in any capacity, “Hot text” is worth reading.

At first “Hot text” threw me off with their hype-filled treatment of object-orientation (as applied to writing, none the less) and XML, but when it got to the meat of their advice on writing, I quickly forgot those initial awkward moments.

The book is written for those who want to improve their writing for the web. It offers both generic tips like:

• writing should be short, simple and clear

• break up long paragraphs into short ones
• make text scannable
• remove hype and marketoid speak
• write in active voice
• write with emotions
• have a personal voice
• link to other sites (gives you credibility)

as well as addresses specific kinds of web content:

• home page
• navigation menus
• help section and FAQs
• product information
• privacy policy
• e-mail responses to customers
• web marketing copy
• press release
• news
• webzines
• e-mail newsletters
• weblogs

While not as deep and inspiring like “On writing well”, it’s filled with practical advice. and definitely worth reading for anyone who has to write for the web (especially those that don’t have prior experience in copy writing).

Jackito is a new PDA by some French startup company. Looks decent but obviously doesn’t have a tiny chance in the crowded marketplace. It doesn’t offer anything substantially better than Palm OS, Pocket PC or Symbian - and those are established players backed by large companies.

What drew my attention was this marketing overview. It says: “The only one with a such larger screen”. While “large” might be interpreted in many ways, it’s screen resolution is only 240×320 - the same as Pocket PC, two times smaller than many Palm PDAs (Tapwave Zodiac, many CLIE models and some Tungsten’s have 320×480), not to mention Toshiba’s Pocket PC with 800×600.

So, lies, all lies, nothing but the lies.

Oh, and according to some Slashdot posts it’s just a thick-plotted scam. They do have a nice website, though. Which prooves that on the Internet, no one knows, that you’re a dog.

• list of coffee shops with free wireless in Seattle
• article on coffe shops with free wifi in Seattle

• SubEthaEdit coder interview
• notcon 2004 audio and video of talks - Brewster Kahle’s talk was great
• outdoor movies in Seattle, summer 2004

p2pq (I think it comes from p2p knowledge) looks like interesting project although I think they’ll fail. Basically they implement “ask the expert” services but instead of using web front-end (as all the existing services do) they use p2p (whatever that means in this context) client.

The problem is: I don’t think this approach scales. Imagine a lot of people start using it. It’s my guess that people asking the questions are also those giving the answer, so basically any question gets broadcasted to everyone on the network, hoping that someone will know the answer. While this mode is crucial for finding answers (the more people on the network, the more likely that someone knows the answer) it’s also floods people with questions that they’re most likely not able to answer. So the system basically chokes itself. If people stop reading (and therefore answering) the questions from other people, the value of the system goes down (you no longer get the answer).

In an ideal world a given question would only be sent to people who know the answer and are willing to type it. However, there’s no technological way to ensure that (absent mind-reading technology).

We can try hacks like categorizing the knowledge e.g. I could say that I only answer questions about wxWindows programming in Python. But categorizing schemes don’t scale either - you have to rely on two random, unconnected parties to agree on categories - ain’t gonna happen.

Besides, there are already plenty of alternatives e.g. if I have a question about wxWindows programming in Python, I will ask it on wxWindows/Python mailing list - that’s the best place to go.