I'm reaching way back in my memory here, but I recall a white paper (perhaps from Usenix) from SGI where they investigated how to keep their scheduler from using so many cycles - not so much from a "improve throughput" thrust, but more so to "improve responsiveness".
Their conclusion was that what you wanted to do was have a two-level scheduler -- a real quick + dirty part that ran at interrupt level and just grabbed the next runnable processes from a circular list of the highest priority processes -- in and out in just a few cycles, but perhaps not grabbing *the* highest priority process this time -- then "every so often" (in computer terms, e.g. some fraction of a second) a lower level scheduler ran which did a more thorough re-ordering of the processes.
Of course, one immediately sees that this lower level scheduler could even be a regular process (making syscalls) which means you can plug in whatever scheduling algorithm you like.
Up to now our annual costs for supplying campus with this kind of bandwidth was been a couple hundred thousand or so, which is low enough that it's a dept budget item and "under the radar" -- but we are projecting our costs will be in the millions/year in the nearish future -- and that number attracts attention from the big suits.
Will schools start to suffer a Soviet-like collapse when they no longer have the resources (or the will to use them) to keep up with other schools?
IIRC, we are already at an OC-48 (48 T3/DS3's) for data/video/phone and OC-192 can't be far -- madness!
I used to consult for a web hosting site and not *1*, that's right NONE of the companies whose web stores they hosted wanted their transactions excrypted -- the time I spent setting up pgp, etc was totally wasted, they all insisted on clear-text e-mail of transactions. And as you know, the customer is always right...
Joe Computer writes: ps. One more slightly paraphrased quote: "I'm sick of working here at The Computer Industry. I'm moving to Iowa to raise corn." If you need me, I'll be in Iowa.
Plenty of us Iowans were stuck on this useless detail too...
And we'd have probably killed for your cold ham and sweedish meatballs...
I wrote a CC verifier several years ago that talked to first data. We asked, they sent a manual detailing the protocol (which I had 99% reverse engineered with a serial analyzer anyway by the time it arrived snail-mail). When some bit was unclear, I called them and they answered my questions on the spot.
I can't speak for anyone else, but I had a very good experience with them.
I realize places like Redmond and SV have largely turned their entire populace into 90-hr/week Stepford Geeks, but elsewhere many of us are just like normal folks (married, kids, houses, soccer practice, etc.) except we work and play with cool stuff. As such I could care less about coffee houses, performance art, Fry's, or hot babes (ok, maybe a little).
Everyone makes fun of Iowa, but I get:
a nice house for ~$100K
a 12 minute commute -- on foot!
our son in a great public school where I can walk with him in the morning
no crime other than drunken college idiots
DSL
good food cheap
friendly people
Of course, Iowa, at least our "leaders", have fallen into the feeling-sorry-for-ourself because 20-somethings want to live in exciting places trap.
Frankly, I think this is stupid -- let'em go -- lure them back when they are settled-down, responsible, law-abiding, tax-paying, 30-something families.
where ${title} and ${body_str} presumably appear in hello_world_blue.html
And the rest of that Java preamble stuff (or the #include... int main... equivalent for C) is nothing more than boilerplate that you either import with a couple of editor keystrokes or just link against. If you keep reinventing that wheel you aren't doing your employer any favors productivity-wise, but in and of itself it's no reason to switch to Perl.
Before the Perl-Patrol goes berserk on me, yes, their are jobs for which Perl is an (perhaps even the) appropriate tool for the job.
Anyway, that example doesn't do justice to all the neat things that Java Servlets do for you, but don't believe me, just read the O'Reilly book on them).
BTW, why the heck can't we use <pre>...</pre> or something! to format code-snippets with reasonable ease!!!!
"the code maintance people will have a hell of a time."
That's part of the problem -- even having "code maintenance people"!
Because, typically, that means the 'hotshots' are all doing feature-creep and the 'code monkeys' are doing maintenance -- is it any wonder we get bloated BSoD crapware as a result?
"To have set up a "unit test" you must already have defined some sort of spec to test against, which isn't what I was talking about"
Actually, in a way the unit test IS the spec.
"Refactoring is a fix-it-after-its-happened strategy. In our company that has turned out to be far more costly than just doing proper planning to begin with."
Yes, it is a fix-after strategy, but it is one I have found really works (even long before I had ever heard of XP). Very often I find myself chopping a function in half or sucking the middle out. It takes me *very* little time to turn:
int foo ( ) { int a1; int a2;
do {something} while (sometest); do {more} while (another); }
into: int foo ( ) { int a1;
do {something} while (sometest); bar(); } int bar ( ) { int a2;
do {more} while (another); }
If I discover I need to use the 'bar' functionality someplace else.
Which is why a successful boycott is the answer. If it is the case that these patents cost more business then they generate it revenue then they have the same legal duty not to patent (or in this case to post facto donate it to the public domain.
Other than the server-to-server communication it is pretty good (esp. considering it is late 80s technology). Of course, if you designed it today it would use MIME and content-type text/html instead of its own simple formatting language @b(this in bold) @i(this in italics) @large(etc).
Anyway, the IETF IMPP mailing lists are hosted here:
impp@iastate.edu . . . (technical stuff) meta-impp@iastate.edu . . . (other stuff) (It's majordomo, you know how to use it)
First, you don't have to throw 10^16 numbers at a CC checker to get a valid one. Look at the first 4 digits, use them. Then make up digits for the rest such that all the digits match the checksum. The checksum is supposed to catch typos and transpositions and such, but it is pretty lame...
Of course, there is NO merit to this stupid patent as anyone who reads the CGI Usenet Newsgroups knows. Even disregarding that software patents are stupid and evil, there is plenty of prior art here -- including some written by me years ago -- and, of course, it's hardly novel, either.
Perhaps having 2 titans go at it will finally shed the needed sunshine on this nonsense -- so perhaps the injunction is a good thing in the end.
I thought TS2 was great (and, of course, our 6 year old loved it). I thought the handling of skin was much improved this time, but I was surprised that they still hadn't gotten realistic walking. Probably the biggest thing left to go is imperfections (i.e., the dust seemed 'flat', the road looked better than any actual road, and where is this town with all shiny cars:) Still, a great story makes it well worth it (plus you get to see a famous short first)
Just how far will our federal government go towards controlling strong encryption? Apparently, very far. And this isn't a new effort by any means. We learned that William Allen Simpson, a Detroit-based computer consultant who was on the IETF staff, has been investigated by the federal government for treason charges. Simpson was the person that argued loudly for encryption to be included in the PPP protocol when it was still in design phases. That push landed Simpson in hot whatever with federal officials. Simpson learned through friends that he was under investigation for treason -- the FBI had been interviewing his friends and associates.
Simpson obtained 54 pages of documents from the government under the Freedom of Information act, however the documents were heavily sensored, including the bureau's basis for the investigation. According to a ZDTV report, Simpson did learn that the FBI had accused him of "challenging authority and laws that may impinge upon his activities
There is a middle ground and that is what we are implementing here -- allowing people to choose what information goes in their (X.509) certificate. So you can get one as generic as "member of the ISU community", or perhaps "Engineering Undergraduate" (say if you needed that to access a certain library journal the Engineering college was paying for [tightwads]), or you could get a specific one (say to check your grades or U-Bill).
Of course, since this is comment 11-million or so, who will read it...
The list is, of course, only as complete as its submissions. For example, we have a 512-cpu Paragon in the basement, that would probably qualify for the list if anyone bothered...
Slashdot Mirror
Their conclusion was that what you wanted to do was have a two-level scheduler -- a real quick + dirty part that ran at interrupt level and just grabbed the next runnable processes from a circular list of the highest priority processes -- in and out in just a few cycles, but perhaps not grabbing *the* highest priority process this time -- then "every so often" (in computer terms, e.g. some fraction of a second) a lower level scheduler ran which did a more thorough re-ordering of the processes.
Of course, one immediately sees that this lower level scheduler could even be a regular process (making syscalls) which means you can plug in whatever scheduling algorithm you like.
Up to now our annual costs for supplying campus with this kind of bandwidth was been a couple hundred thousand or so, which is low enough that it's a dept budget item and "under the radar" -- but we are projecting our costs will be in the millions/year in the nearish future -- and that number attracts attention from the big suits.
Will schools start to suffer a Soviet-like collapse when they no longer have the resources (or the will to use them) to keep up with other schools?
IIRC, we are already at an OC-48 (48 T3/DS3's) for data/video/phone and OC-192 can't be far -- madness!
I used to consult for a web hosting site and not *1*, that's right NONE of the companies whose web stores they hosted wanted their transactions excrypted -- the time I spent setting up pgp, etc was totally wasted, they all insisted on clear-text e-mail of transactions. And as you know, the customer is always right...
ps. One more slightly paraphrased quote: "I'm sick of working here at The Computer Industry. I'm moving to Iowa to raise corn." If you need me, I'll be in Iowa.
Plenty of us Iowans were stuck on this useless detail too...
And we'd have probably killed for your cold ham and sweedish meatballs...
I can't speak for anyone else, but I had a very good experience with them.
PS, we'll sell for less than $80M :)
Everyone makes fun of Iowa, but I get:
a nice house for ~$100K
a 12 minute commute -- on foot!
our son in a great public school where I can walk with him in the morning
no crime other than drunken college idiots
DSL
good food cheap
friendly people
Of course, Iowa, at least our "leaders", have fallen into the feeling-sorry-for-ourself because 20-somethings want to live in exciting places trap.
Frankly, I think this is stupid -- let'em go -- lure them back when they are settled-down, responsible, law-abiding, tax-paying, 30-something families.
subsVar("title", "Hello World");
subsVar("body_str", "Hellow World!");
subsOut(stdout, "hello_world_blue.html");
where ${title} and ${body_str} presumably appear in hello_world_blue.html
And the rest of that Java preamble stuff (or the #include ... int main... equivalent for C) is nothing more than boilerplate that you either import with a couple of editor keystrokes or just link against. If you keep reinventing that wheel you aren't doing your employer any favors productivity-wise, but in and of itself it's no reason to switch to Perl.
Before the Perl-Patrol goes berserk on me, yes, their are jobs for which Perl is an (perhaps even the) appropriate tool for the job.
Anyway, that example doesn't do justice to all the neat things that Java Servlets do for you, but don't believe me, just read the O'Reilly book on them).
BTW, why the heck can't we use <pre>...</pre> or something! to format code-snippets with reasonable ease!!!!
Serbian Nationalist Gavrilo Princip started WWI when he assassinated Archduke Ferdinand in Sarajevo
That's part of the problem -- even having "code maintenance people"!
Because, typically, that means the 'hotshots' are all doing feature-creep and the 'code monkeys' are doing maintenance -- is it any wonder we get bloated BSoD crapware as a result?
Actually, in a way the unit test IS the spec.
"Refactoring is a fix-it-after-its-happened strategy. In our company that has turned out to be far more costly than just doing proper planning to begin with."
Yes, it is a fix-after strategy, but it is one I have found really works (even long before I had ever heard of XP). Very often I find myself chopping a function in half or sucking the middle out. It takes me *very* little time to turn:
int foo (
) {
int a1;
int a2;
do {something} while (sometest);
do {more} while (another);
}
into:
int foo (
) {
int a1;
do {something} while (sometest);
bar();
}
int bar (
) {
int a2;
do {more} while (another);
}
If I discover I need to use the 'bar' functionality someplace else.
Other than the server-to-server communication it is pretty good (esp. considering it is late 80s technology). Of course, if you designed it today it would use MIME and content-type text/html instead of its own simple formatting language @b(this in bold) @i(this in italics) @large(etc).
Anyway, the IETF IMPP mailing lists are hosted here:
impp@iastate.edu . . . (technical stuff)
meta-impp@iastate.edu . . . (other stuff)
(It's majordomo, you know how to use it)
- IBM created SQL, but
- SQL is not a good thing created by IBM.
It is grossly inferior to QUEL.#include<stdlib.h>
intluhn_ok(
char*inp
){
return((luhn(inp)%10) ==0);
}
intluhn(
char*inp
){
staticint x[2][10]= {0,1,2,3,4,5,6,7,8,9,0,2,4,6,8,1,3,5,7,9};
char* p;
int s =0;
int sum =0;
char c;
if((inp==NULL)|| (*inp=='\0'))return -1;/*biteme, doughboy!*/
for(p=inp;*p !='\0';++p){}
do{
c=*--p;
if((c<'0')|| (c>'9'))continue;
sum+=x[s][c-'0'];
s^=1;
}while(p!=inp);
returnsum;
}
Compared to the now-vanquished QUEL from ingres it's pretty weak.
And, of course, it has an icky COBOL-like look+feel.
All, in all, it's the EBCDIC of DB languages.
Perhaps having 2 titans go at it will finally shed the needed sunshine on this nonsense -- so perhaps the injunction is a good thing in the end.
I thought TS2 was great (and, of course, our 6 year old loved it). I thought the handling of skin was much improved this time, but I was surprised that they still hadn't gotten realistic walking. Probably the biggest thing left to go is imperfections (i.e., the dust seemed 'flat', the road looked better than any actual road, and where is this town with all shiny cars :) Still, a great story makes it well worth it (plus you get to see a famous short first)
Of course, since this is comment 11-million or so, who will read it...
The list is, of course, only as complete as its submissions. For example, we have a 512-cpu Paragon in the basement, that would probably qualify for the list if anyone bothered...