1) solaris SMP is a 3rd rate hack, its been proven by alan cox
Do you have any links or more information about Solaris's "third rate" SMP. I was under the (marketing?) impression that Solaris had kick-ass SMP. I'm particularly curious as to what Alan Cox had to Solaris's SMP. I've been following the linux-kernel mailing list and I'm very excited about Linux 2.4's SMP improvements.
What's the answer for those of us without HPUX?
on
Gartner Slams Linux
·
· Score: 1
There is plenty of "winning" IBM technology here..:-)
I love how Mountain Dew has almost no carbonation!
on
MTV Profiles "Hackers"
·
· Score: 2
I've been a major dew-aholic for years. It's so smooth and chuggable. I can't drink Coke or Pepsi. They are so heavily carbonated that you end up drinking a mouthfull of froth. yew..
The Debian distro is very conservative. They've been heavily testing Debian "potato" for a long time. I doubt they would suddenly through an untested new 2.4 kernel into their distro. They would have to restart all of their testing.
You might not be able to click the OK button offscreen, but you can use the cursors or ESC. I've "driven blindly" when repairing many hosed NT machines.. sigh..
Writing programs that rely on Windows DLLs is like a building a house of cards. Here's a great DDJ article about how to solve your DLL woes - past, present, and future!
The answer to the almost limitless problems of DLLs is obvious: Don't use them. Wherever possible, use static linking. Imagine the benefits. Some other developer's boneheaded installation or poorly designed updated DLL will not break your application. Your application won't fail because a component is missing, or because a registry setting has been lost or modified incorrectly. Your application won't behave differently depending on the applications already loaded, as a DLL-based application can if another application has already loaded a different copy of one of its components. Your installation will be exceptionally simple, and an uninstall will be just as easy.
I should warn you of one small hassle if you try DLL-free development. Your users won't believe that there is only one file to install.
Drudd's post is right on. Comctl32.dll is a great example. Windows 2000 includes a new version of the common controls, called Comctl20.dll. For backwards compatibility, it also includes a Comctl32.dll. Unfortunately, the Comctl32.dll in Windows 2000 is a different Comctl32.dll than other versions of Windows! Windows 2000's Comctl32.dll is a simply a wrapper DLL that thunks API calls up to the new Comctl20.dll.
As a third-party developer, if you link your program with Comctl32.dll on NT4 or Windows 9x, you must code around known Comctl32.dll bugs. When users upgrade to Windows 2000, your program dynamically links with the "improved" Comctl32.dll (which calls buggy new code in Comctl20.dll). Your tested program suddenly inherits new bugs from Comctl20.dll. Plus, your work-arounds for known bugs in Comctl32.dll might have bad side effects when the bugs disappeared in the Comctl20.dll.
This is a terrible design decision and typical of Microsoft's "engineering". On Linux, a library developer would probably ship the well-known old.so and the new.so with different filename reflecting the library version numbers. Remember the problems of some Linux distros mixing up libc5 and libc6? Microsoft does this all the time..
"Amazon.com has a market capitalization of $5.75 billion (August 10, 1998). They built their site with compiled C CGI scripts connecting to a relational database. You could not pick a tool with a less convenient development cycle. You could not pick a tool with lower performance (forking CGI then opening a connection to the RDBMS). They worked around the slow development cycle by hiring very talented programmers. They worked around the inefficiencies of CGI by purchasing massive Unix boxes ten times larger than necessary. Wasteful? Sure. But insignificant compared to the value of the company that they built by focusing on the application and not fighting bugs in some award-winning Web connectivity tool programmed by idiots and tested by no one."
I am curious if anyone here has experience implementing DB2 or Oracle on Linux in a "production environment". I'm especially interested in people moving from other operating systems to Linux. What software and people problems did you have? Did you have any difficulty getting customer support for Linux compared to the same database support on commercial operating systems?
From what I've read, IBM seems to be buying into Linux more than Oracle. Oracle seems to see Linux as simply a jab at Microsoft.
Are you talking about "Dr Watson" ? :-)
on
Linux Turns 8
·
· Score: 2
They have obtained a few patents, which contain technical details. Go to www.patents.ibm.com and search for "transmeta".
US05926832 07/20/1999 Method and apparatus for aliasing memory data in an advanced microprocessor
US05905855 05/18/1999 Method and apparatus for correcting errors in computer systems
US05832205 11/03/1998 Memory controller for a microprocessor for detecting a failure of speculation on the physical nature of a component being addressed
Linux gets the attention of the press and Big Companies, so most commercial applications seem to be showing up on Linux before FreeBSD. How risky is running a big app like Oracle 8i or IBM DB2 using FreeBSD's Linux "emulation"? Any problems?
If you have to look at banner ads, wouldn't you rather see ads that you actually might be interested in? Ads are annoying, but if they were actually targetted to me, they might prove worthwhile occasionally. Of course, this ignores any privacy concerns regarding Doubleclick's user profiles.
Not a bad idea!;-) A new internet startup called Akamai does something very similar. The company has ~900 servers around the world, caching their clients static web content. For example, jcrew.com generates their HTML with image URLs pointing to (say) http://a1240.g.akamaitech.net/7/1240/969/ffe0a8c13 22031/jcrew.com/images/ sep99/e2home/clearance.jpg. This distributes the load off of jcrew.com's server to the Akamai servers nearest the user.
I wonder if someone could write a freebie script that does the same thing using free web space like GeoCities or Xoom. The script could automatically create accounts on those sites and shed the load for serving static content to these free servers. >:-) I bet GeoCities would soon figure out a way to block this behaviour, though.. but it still might be fun.
True, I have a slow connection. Funny, though, how other sites like Yahoo or photo.net have snappy performance on the same Internet connection..:-)
Please see Jacob Nielsen's The Need for Speed. As the Internet grows, the mean bandwidth per user decreases. Most new users are using modems, not T-1s. Web sites should design for the common case and minimize response time.
Even so, my browser oftens prints the Slashdot menu and sidebars, but hangs for a looong time waiting for the long list of messages. My browser has to wait for all (say) 200 message titles to download before printing even the first title. If the message list used no tables, the first title could be printed immediately. The message list could even be broken down into multiple tables, stacking them vertically. That way, my browser could print the title list in chunks of (say) 20 titles per table. Of course, this would need to be benchmarked.:-)
Slashdot uses HTML tables to organize the web page "newspaper-like". The problem with HTML tables is the browser cannot draw their contents until all of the table data arrives at the browser. With a dynamic site like Slashdot, some data might take longer to pull out of the database than others. If Slashdot used fewer tables, then at least a portion of the page could be rendered. The users could read the top half of the page while waiting for Slashdot to generate the second half of the page.
If your data is so important that you need RAID 5 on your servers, why are you using MySQL, which does not support database transactions? I admit I have not used MySQL, but the impression I got was that it was a little "fast and loose" with regards to reliability. Did you consider any databases that support transactions like PostgreSQL or Oracle? You seem to have really done your homework and I'm very interested to read what you've found out!
If Apache is using mod_perl to serve dynamic content, how does Apache pass the page to Squid? Does the original HTTP request go to Squid on port 80 and then Squid hands the request off to Apache? Should Squid be on a firewall-like machine in front of the Apache machine?
User computer --> Squid server --> Apache server --> DB server?
Companies like Microsoft and Sun spend millions of dollars on software testing. Can the open source community really replace all those professional software testers? What tools or techniques do you personally use to test your code? What programs do you use to "stress test" Linux?
This explains why Passport is so slow and broken.
on
Hotmail Cracked Badly
·
· Score: 1
I use two MSN sites that use Microsoft Passport, Hotmail and MSN Investor. They refuse to cooperate with Passport! Investor has a feature to store your portfolio on a centalized server so you can view it from any web browser (after authentication), but that portfolio never responds or it scrambles my portfolio data. When I then jump to Hotmail, it forgets my password (which I asked it to remember on my home computer). Damn this software is stinky..
1) solaris SMP is a 3rd rate hack, its been proven by alan cox
Do you have any links or more information about Solaris's "third rate" SMP. I was under the (marketing?) impression that Solaris had kick-ass SMP. I'm particularly curious as to what Alan Cox had to Solaris's SMP. I've been following the linux-kernel mailing list and I'm very excited about Linux 2.4's SMP improvements.
There is plenty of "winning" IBM technology here.. :-)
I've been a major dew-aholic for years. It's so smooth and chuggable. I can't drink Coke or Pepsi. They are so heavily carbonated that you end up drinking a mouthfull of froth. yew..
Here's info straight from the Debian web site:
Major Changes For "potato":
The major changes we know will be part of potato are the following:
Linux kernel 2.2 based (for architectures with Linux kernels at all).
GNU libc 2.1 based (upwardly binary compatible with GNU libc 2.0)
The egcs version of gcc, also known as egcc, will be the default C compiler.
There is also a good chance that the potato release will include the powerpc and arm architectures.
You might not be able to click the OK button offscreen, but you can use the cursors or ESC. I've "driven blindly" when repairing many hosed NT machines.. sigh..
Oops, one more comment:
Writing programs that rely on Windows DLLs is like a building a house of cards. Here's a great DDJ article about how to solve your DLL woes - past, present, and future!
"Windows DLLs: Threat or Menace?"
Here's an excert:
The answer to the almost limitless problems of DLLs is obvious: Don't use them. Wherever possible, use static linking. Imagine the benefits. Some other developer's boneheaded installation or poorly designed updated DLL will not break your application. Your application won't fail because a component is missing, or because a registry setting has been lost or modified incorrectly. Your application won't behave differently depending on the applications already loaded, as a DLL-based application can if another application has already loaded a different copy of one of its components. Your installation will be exceptionally simple, and an uninstall will be just as easy.
I should warn you of one small hassle if you try DLL-free development. Your users won't believe that there is only one file to install.
Drudd's post is right on. Comctl32.dll is a great example. Windows 2000 includes a new version of the common controls, called Comctl20.dll. For backwards compatibility, it also includes a Comctl32.dll. Unfortunately, the Comctl32.dll in Windows 2000 is a different Comctl32.dll than other versions of Windows! Windows 2000's Comctl32.dll is a simply a wrapper DLL that thunks API calls up to the new Comctl20.dll.
.so and the new .so with different filename reflecting the library version numbers. Remember the problems of some Linux distros mixing up libc5 and libc6? Microsoft does this all the time..
As a third-party developer, if you link your program with Comctl32.dll on NT4 or Windows 9x, you must code around known Comctl32.dll bugs. When users upgrade to Windows 2000, your program dynamically links with the "improved" Comctl32.dll (which calls buggy new code in Comctl20.dll). Your tested program suddenly inherits new bugs from Comctl20.dll. Plus, your work-arounds for known bugs in Comctl32.dll might have bad side effects when the bugs disappeared in the Comctl20.dll.
This is a terrible design decision and typical of Microsoft's "engineering". On Linux, a library developer would probably ship the well-known old
"Amazon.com has a market capitalization of $5.75 billion (August 10, 1998). They built their site with compiled C CGI scripts connecting to a relational database. You could not pick a tool with a less convenient development cycle. You could not pick a tool with lower performance (forking CGI then opening a connection to the RDBMS). They worked around the slow development cycle by hiring very talented programmers. They worked around the inefficiencies of CGI by purchasing massive Unix boxes ten times larger than necessary. Wasteful? Sure. But insignificant compared to the value of the company that they built by focusing on the application and not fighting bugs in some award-winning Web connectivity tool programmed by idiots and tested by no one."
I am curious if anyone here has experience implementing DB2 or Oracle on Linux in a "production environment". I'm especially interested in people moving from other operating systems to Linux. What software and people problems did you have? Did you have any difficulty getting customer support for Linux compared to the same database support on commercial operating systems?
From what I've read, IBM seems to be buying into Linux more than Oracle. Oracle seems to see Linux as simply a jab at Microsoft.
C:\WINNT\DRWTSN32.EXE
Linux gets the attention of the press and Big Companies, so most commercial applications seem to be showing up on Linux before FreeBSD. How risky is running a big app like Oracle 8i or IBM DB2 using FreeBSD's Linux "emulation"? Any problems?
What are these 1x1 pixel GIFs used for? Ad tracking, I presume?
If you have to look at banner ads, wouldn't you rather see ads that you actually might be interested in? Ads are annoying, but if they were actually targetted to me, they might prove worthwhile occasionally. Of course, this ignores any privacy concerns regarding Doubleclick's user profiles.
Not a bad idea! ;-) A new internet startup called Akamai does something very similar. The company has ~900 servers around the world, caching their clients static web content. For example, jcrew.com generates their HTML with image URLs pointing to (say) http://a1240.g.akamaitech.net/7/1240/969/ffe0a8c13 22031/jcrew.com/images/ sep99/e2home/clearance.jpg. This distributes the load off of jcrew.com's server to the Akamai servers nearest the user.
I wonder if someone could write a freebie script that does the same thing using free web space like GeoCities or Xoom. The script could automatically create accounts on those sites and shed the load for serving static content to these free servers. >:-) I bet GeoCities would soon figure out a way to block this behaviour, though.. but it still might be fun.
True, I have a slow connection. Funny, though, how other sites like Yahoo or photo.net have snappy performance on the same Internet connection.. :-)
Please see Jacob Nielsen's The Need for Speed. As the Internet grows, the mean bandwidth per user decreases. Most new users are using modems, not T-1s. Web sites should design for the common case and minimize response time.
Even so, my browser oftens prints the Slashdot menu and sidebars, but hangs for a looong time waiting for the long list of messages. My browser has to wait for all (say) 200 message titles to download before printing even the first title. If the message list used no tables, the first title could be printed immediately. The message list could even be broken down into multiple tables, stacking them vertically. That way, my browser could print the title list in chunks of (say) 20 titles per table. Of course, this would need to be benchmarked. :-)
Is Linux 2.0.x dead? Will you keep supporting it? Will there be a Linux 2.0.39?
Slashdot uses HTML tables to organize the web page "newspaper-like". The problem with HTML tables is the browser cannot draw their contents until all of the table data arrives at the browser. With a dynamic site like Slashdot, some data might take longer to pull out of the database than others. If Slashdot used fewer tables, then at least a portion of the page could be rendered. The users could read the top half of the page while waiting for Slashdot to generate the second half of the page.
If your data is so important that you need RAID 5 on your servers, why are you using MySQL, which does not support database transactions? I admit I have not used MySQL, but the impression I got was that it was a little "fast and loose" with regards to reliability. Did you consider any databases that support transactions like PostgreSQL or Oracle? You seem to have really done your homework and I'm very interested to read what you've found out!
If Apache is using mod_perl to serve dynamic content, how does Apache pass the page to Squid? Does the original HTTP request go to Squid on port 80 and then Squid hands the request off to Apache? Should Squid be on a firewall-like machine in front of the Apache machine?
User computer --> Squid server --> Apache server --> DB server?
Companies like Microsoft and Sun spend millions of dollars on software testing. Can the open source community really replace all those professional software testers? What tools or techniques do you personally use to test your code? What programs do you use to "stress test" Linux?
I use two MSN sites that use Microsoft Passport, Hotmail and MSN Investor. They refuse to cooperate with Passport! Investor has a feature to store your portfolio on a centalized server so you can view it from any web browser (after authentication), but that portfolio never responds or it scrambles my portfolio data. When I then jump to Hotmail, it forgets my password (which I asked it to remember on my home computer). Damn this software is stinky..