|
|
Tuesday, February 25, 2003 |
Yes, I'm at home today because of bad weather. Yes, I'm a posting fool today. Yes, I'm waiting for large downloads over an ADSL connection...
11:23:49 AM
|
|
The Inquirer: Trusted Computing Platform becomes real in Intel Springdale. [Hack the Planet]
Here comes Trusted Computing. Intel is set to release Springdale in May. I have mixed feelings about this. It sounds like a good idea - making sure software delivery is from trusted sources and helping users prevent software failure. However, I'm concerned about potential privacy issues and platform lock-in.
11:22:54 AM
|
|
"For each language, student researcher Alexandra Rahlin measured the execution time and the amount of virtual memory allocated to the task. She deemed the fastest language using the least memory "most efficient." "
I don't want to take anything away from Ms. Rahlin's research, but I think everyone expected this, right? Another good source of language performance is The Great Computer Language Shootout.
10:58:05 AM
|
|
Dancing with the devil.
Blue Screen of Death |
| While trying to stabilize a Windows server yesterday, Dave Winer pointed to a symphony of BSODs and wrote:
Windows itself is stable, but the server junk they throw on top of it blue screens, probably because MS engineers don't feel that they have to follow the rules, and they have access to the OS source code. Scripting News
My impression is that it's not access to source code that causes these problems, but rather, dancing with the devil at ring 0. Anybody who writes a device driver has to do that dance. Of course, the increasing kernelization of servers, for performance, does make them more device-driver-like. In Windows Server 2003, for example, the HTTP stack moves into the kernel -- a prospect that is both exciting and scary. ... [Jon's Radio]
I remember when I saw a Byte (?) article about the "improvements" Micrsoft made in Windows NT 4. That was when MS started moving large amounts of code into the kernel layer. I think that's when they started moving some of the GUI stuff into the kernel layer in the name of performance. If you remember NT 3.51, you'll remember that there were some SERIOUS performance problems, especially around the user interface.
I bring this up quite frequently with my other geeky friends when talking about Windows stability and why it is so awful. MS made the decision to violate some of the principals of operating system design that alot of other people try to adhere to. When they did that, they sacrificed stability.
Now, it looks as though everything is pushing into the sacred domain of the kernel. This seems like a VERY bad idea. It's bad enough that you have to do that for lower priority pieces of the operating system. But, when you start doing it for user space programs, it seems like a VERY BAD idea.
9:49:48 AM
|
|
Trust and knowledge sharing. The IBM Institute for Knowledge-based Organizations has released the results of recent research into trust and knowledge sharing. To quote: Recently, the IBM Institute for Knowledge-Based Organizations (IKO) studied the role of trust in knowledge sharing. Factors such as the... [Column Two]
9:31:18 AM
|
|
Employee directory search: one box or two?. Jakob Nielsen's latest Alertbox discusses interface design issues around employee directory search. To quote: Guidelines conflict on whether to limit intranet search to a single search box or dedicate an additional box to employee directory searches. There's theory to support... [Column Two]
We're getting ready to ramp up our internal search capabilities at the company I work for. This might be an interesting read releated to that.
9:27:33 AM
|
|
New site: Contentology. I've just been made aware of a new website: Contentology, which provides an interesting mix of resources in the areas of content management and associated disciplines. To quote: Contentology blends disciplines such as information architecture, information design, knowledge management, communications... [Column Two]
9:25:32 AM
|
|
I finally read "The Rise of 'Worse is Better'" this morning. This is a short essay on two approaches to software design - "the right thing" and "worse-is-better". I've seen references to it from time to time, but never actually read it. It's entertaining and uses humor to underscore an important point. Sometimes it is important to do "the right thing" and sometimes it is important to take a more simplistic approach and go for ease of implementation.
I have a tendency to lean toward "the right thing" approach and less towards "worse-is-better". I can recognize times that I should have sacrificed some elegance and/or completeness for ease of implementation. I justify it by imagining that my design is easier to extend or maintain. However, I jeopardize project schedules or budget at time when I get hung up on the 20% that takes 80% of the time.
There is an important and delicate balance between "the right thing" and "worse-is-better" that all consultants face. I believe that consultants MUST leave behind code that is maintainable and understandable. Consultants MUST do as much of "the right thing" as possible without emptying the client's pocketbook. And, finally, consultants MUST be capable of doing "the right thing" on a tight budget because that's one reason the clients hire us.
9:17:59 AM
|
|
© Copyright 2004 Tom Pierce.
|
|
|
|
|
Search
|
Emacs Sources
tsql-indent.el
This is an indentation function for SQL mode. It was written with Transact SQL in mind.
user-add-sql-folding-marks
This is a simple function that adds folding-mode marks to SQL sources. It is quick and dirty, but is fairly useful for me.
remove-line-boundary-in-region
This function removes all the line boundaries in a region. This, in effect, collapses all the lines in the region onto one line.
convert-camel-to-underscore
This function converts all the text that is camel cased in a particular region to underscore separated text.
My Subscriptions
|
|
|