Monday, August 26, 2002

Forgot to mention I forgot to mention on the Google thing. The built in Internet search channel that comes with Sherlock actually has code to search Google, but it's commented out. 3:59:15 PM    comment Google Search Channel for Sherlock 3.0 While I've been hacking around with Sherlock I managed to put together a new search channel to search Google. In every sense of the word this is a hack since Apple hasn't released the Sherlock SDK yet. The main problem is altering the UI elements. Without the SDK you can't open the nib files in Interface Builder so I used the AppleCare channel UI and hacked it with a hex editor to change some of the strings. The scripts are easy once you battle through the weird conglomeration of Javascript and XQuery that they are using. Anyway, to use the channel it's really easy. Here is the Sherlock URL to add this channel to Sherlock. sherlock://www.xmldatabases.org/projects/GoogleSearch/Channel.xml?action=add In Internet Explorer you should just be able to click the link and it will automatically add it. Other browsers may not work, but you should be able to add it under Sherlock : Preferences : Subscriptions, click Add and then use this URL. http://www.xmldatabases.org/projects/GoogleSearch/channels.xml This channel is an experiment and really only exists because I wanted to figure out how Sherlock works. Once Apple releases the SDK it will be possible to build much more compelling examples. BTW, if anyone knows where I can get the SDK please let me know. What this does is search Google and return the first 100 results. Clicking the search button again will return another 100 results and so on. Single clicking an item shows the abstract in the bottom window and double clicking it opens it in a new browser window. There are a few caveats with this. The relevance column isn't used, it's just there because it's part of the AppleCare interface and I couldn't remove it. Search results aren't returned in the same order that Google returns them, I'm trying to figure this out. Now the interesting part about how Sherlock works is that if I update the channel it should be updated automatically in Sherlock. This provides a way to offer limited functionality applications with native Mac OS X GUIs and strong web services support via a centrally managed mechanism. Other then the Google pieces, all the logic for the channel runs within a constrained environment within Sherlock. You don't have full native access to the local PC since it would be horribly insecure. There is much more that can be done here then what Apple is currently using. In environments where Mac OS X is the main OS, Sherlock should become a tool for many things well beyond toy apps like searching Google. I'll try to explain more on what I mean in the future. 3:50:17 PM    comment Changing the Zen of Programming Changing the Zen of Programming. I can remember when a program was developed as a single, self-contained and monolithic piece of code. But the way we program is changing. Our tools are changing, our requirements are changing, and the environments in which our programs are deployed are changing. [osOpinion] This is something that programmers really need to accept. The concept of managed code (as Microsoft calls it) is here to stay and within the next ten maybe even five years will be by far the most dominant way of developing software. The way software is developed right now sucks and it's the programmers who are keeping it that way by trying to "stick to the metal" in the name of performance. We need to get away from this way of thinking and accept the next level of abstraction as the norm. Writing OS kernels in C is one thing, but believing that is the way forward for writing end user applications is a dead end. 1:22:16 PM    comment RSS and UDDI Karsten Januszewski from Microsoft floats a trial balloon for using UDDI to locate RSS files. [Scripting News] Wow talk about beating a dead horse. RSS is popular because it's simple and because it's simple it's easy for people to understand and put to work. UDDI isn't any of those things and because of that has no place around RSS. In fact UDDI really just needs a bullet to the head to put it to rest. Want a useful way to discover RSS feeds? Just put a HTML page on your site, say here is my RSS feed and I like to talk about XYZ. Make sure the page is accessible to Google and a thousand times more people will be able to find your feed then if it's in any stupid UDDI repository. Want programmatic discovery? Somebody can write a 20 line spec that provides a template for the page and use Google as a web service. Make it simple enough and people will do it just because it only takes five minutes to put together. It takes more the five minutes just to figure out what UDDI is supposed to do. 12:52:19 PM    comment

© Copyright 2002 Kimbro Staken. Last update: 8/26/02; 3:59:40 PM. This theme is based on the SoundWaves (blue) Manila theme.