Have you tried kernel 2.6.0? It supposedly improves Linux's responsiveness by quite a bit.
My understanding is that 2.6 improves responsiveness (latency) and, quite inevitably, sacrifices raw throughput. Application load time is all about throughput, so it's probably not going to be the case that dropping in a 2.6 kernel makes applications start faster. It seems to me that this will need to be addressed at higher layers.
I don't understand why Mozilla has ANY form of disk caching built in in the first place - that is not the way of Unix.
You might not know this, but Mozilla doesn't just run on Unices. It also runs on Windows, Mac OS, and god knows what else. Most of what Mozilla does is not the way of Unix, mainly for the sake of being cross-platform.
Anyway, if you're interested in "small, sharp tools" or "one job, one program," you should look at Firebird and Thunderbird. You might be interested in knowing that this is the direction Mozilla is heading. So the Moz dev team would appear to agree with you there -- don't hold your breath about losing the disk cache, though.
a) Linux has been built from contributions by numerous unrelated and unknown software developers
b)Many Linux contributors were originally UNIX developers
It's certainly misleading, but it's not a contradiction.
He didn't say Linux was built entirely by unrelated and unknown developers. He said numerous. Therefore this allows for many Linux contributors to be UNIX developers.
Of course, the letter certainly is riddled with errors and outright lies, but this particular one is not a (logical) contradiction.
10) Go to their website, the educational discount pricing was released a few days ago. Or is their website not public enough?
Care to point out where? Because I called Red Hat sales and was told to keep checking the website over the next couple of weeks as they would have some sort of formal announcement there.
Indeed, all the questions about educational discounts were seriously dodged in this interview. There were 3 separate questions posted and basically a one-liner by Szulik saying "Hey $50k sounds good, give me a call and we'll take your money" without any regard to the real questions.
As a heavy but non-technical computer user it has been extremely frustrating for me to encounter 404 errors.
I have noticed, during all this Verisign SiteFinder nonsense, that many people who should know better don't seem to have any idea what a 404 error is. (Not referring to the parent poster, of course, but many others.)
404 isn't what you get when you type in a non-existent domain. That depends on your browser. Error 404 is an HTTP status code that a real live webserver returns when you request a document it can't find.
The only way you're going to get a 404 error is if you type in a VALID domain, and so 404's have nothing to do with Site Finder at all.
I won't be doing any future business with Verisign, and I plan to transfer my domains to another registrar.
I never much liked Verisign in the past, but since I already had an account there, using them to register new domains was simply the path of least resistance. But their SiteFinder is the straw that broke the camel's back.
Or you could compress it, but then for people with home theaters this sucks in quality.
I have encoded and compressed my entire DVD collection into AVI (mpeg4). I don't compress the audio, so of course I get the same AC3 stream I get when I watch from the DVD, and with constant quality encoding, the video quality is such that I usually can't tell the difference. (Of course mpeg4 has its weak points, like dark scenes, or foggy/smoky scenes.)
The disavantage of this is that I don't get DVD menus and special features, but if the movie is all you're interested in, you can squeeze it down into about 1.5G (give or take) and your common movie enthusiast (of which I am one, with a home theatre and all) would find perfectly acceptable in quality.
Well, correct me if I'm wrong, but Ethernet has a standard for how many segments you can tack together (5 is it?)
Adding a router does not extend the segment. It creates a new segment and a new subnet. The 5-4-3 rule does not apply to routers. Just imagine how broken the Internet would be if we could have at most 4 routers between end points.:)
Jason.
Re:must be running the server over Bongo
on
TCP/IP over Bongo Drums
·
· Score: 4, Insightful
Yes, being the net admin at that site, I really wished the submitter of this story would have given me a day or two warning before submitting this story to slashdot.
Well, I've increased the upstream bandwidth allowed for that box, and upped the max number of http processes. *fingers crossed*
Post-it notes by keyboards don't bother me so much, unless they are on mission-critical accounts,...
But didn't you just say:
Users simply do not understand why passwords are important. They are completely unaware of the concept of a bad password (say, "apple") being cracked by a dictionary attack, and then being used as a stepping stone to gain root (at which point it's all over).
A janitor sticking his mail password on his monitor might not be so disastrous, but, as you say, stepping stone...
If people want to write their passwords in their wallet (with no reference to which account it is), because, well, to borrow Schneier's (paraphrased) words, "I have a life time of experience keeping my wallet safe." But post-it notes by the keyboard is definitely where I draw the line -- even if it is only a janitor.
You might want to look at Columba, which basically appears to be an Evolution clone (mail part only) written in Java.
I'm a very happy Evo user, but I did check out Columba just to see what it was all about. It's a bit rough around the edges, but it was actually very functional, and I was impressed at its speed, considering it's Java.
In related news, Microsoft, also well-known for their Rolling Stones Start Me Up campaign in '95, have signed a deal with the Olsen Twins and will begin filming their new movie, "Passport to Microsoft," expected to be released summer of 2004.
Additionally, Python is an embodiment of Open Source, because the code is actually readable and concise enough to lower the barrier of reading it.
At the risk of being redundant, I have to emphatically agree with this. A few years ago I started a project that required me to wrap a C library as a python module. (The project was ORBit-Python.) Having done a lot in perlXS before that, I was quite prepared to struggle with the Python/C API.
But it wound up being truly a breath of fresh air. There are a few sticky points to get hung up on, like what functions return borrowed or new references, but the syntax is elegant and consistent, and the Python code itself is completely intuitive and a pleasure to read.
My head just exploded trying to figure that out. Damnit, what a mess.
Re:OT: I bet your "realtime" MPEG is nowhere near.
on
Jaguar is Over
·
· Score: 1
You seem to have lost the key point-- "without noticable artifacts".
I haven't lost anything. No such point was made to me.
Does your *realtime* MPEG4 encoder do realtime, without noticible artifacts? I think not.
I wasn't aware there is an option for lossless compression specified in MPEG4. If that's the case (and I don't believe it is, but I could be wrong), then you're certainly right. But assuming MPEG4 is, as I believe it to be, strictly a lossy codec, then you're not going to get around artifacting in your video without applying smoothening filters that would only further separate the compressed video from its original. That's the sacrifice you make with lossy codecs.
I don't care to argue about processor specifics. It's completely irrelevant. I simply posted to refute the implied claim of parent poster that x86 is incapable of doing MPEG4 in real-time.
I use mencoder (mplayer) to do my encoding. I use CBR because it's slightly less CPU intensive. CBR is, by its very nature, more likely to produce compression artifacting. I can do VBR (constant quantizing) in real-time, but I prefer to have the wiggle room in CPU usage, and since the video source is television, I don't notice the quality difference. But I have not seen MPEG4 at sane bitrates (less than 2500kbit, say) that did not yield perceptible artifacting. I'd certainly be interested in seeing your examples.
Cheers, Jason.
Re:This will be another solid update
on
Jaguar is Over
·
· Score: 2, Informative
My little iMac can encode MPEG4 video in realtime. Show me an x86 that can do that. Or, shut up about x86 performance.
My x86 PC does that trivially. It's a 2-year-old Athlon 1400. I record TV realtime in mpeg4 (2500kbit) and mp3 (160kbit) with 30-40% CPU to spare.
Jason.
Re:Do younger minds absorb quicker?
on
Ageism in IT?
·
· Score: 1
And why would you care about a degree and not a diploma? Let me guess: it's because YOU have a degree and not a diploma?
At least in Canada, there are distinct differences between the education you get behind diplomas and degrees. Depending on the position for which we are hiring, we may prefer someone with a stronger academic and theoeretical background, because it's been my experience that someone with a good education in theory picks up (and more importantly, understands) practices a lot better than someone with a diploma, and hence a solid practical knowledge, picks up the theory.
It's easy enough to go through the motions for installing software, or adding a user, or tossing in an ACL on a Cisco router. But when it comes time to actually design software or design a network, I'll tend to prefer someone with a degree who has taken courses in software engineering and computational theory.
My understanding is that things are a little different in the US, if that's where you happen to live. I believe there are fewer strict requirements on what a school is able to call a degree or a diploma, whereas in Canada those things are a little more clearly defined.
Jason.
Re:Do younger minds absorb quicker?
on
Ageism in IT?
·
· Score: 1
Actually, any company that would toss a resume because it didn't list college experience isn't a place I'd want to work anyway.
But when it comes down to roughly equivalent work experience, a degree (not a diploma, I don't care about diplomas) will tip the balance, even if the applicant without a degree might look slightly better on paper. When it comes down to it, you have to pick a short list, and whether or not the applicant has a degree is a good metric to use.
Of course I understand what you're saying. Tossing a resume of someone with 40 years of experience but no degree when he's up against a recent graduate is ludicrous. Still, by and large, it's been my experience that people with degrees (which, when you're reading a resume, is about as good an indication as you can get about someone's academic and theoretical background) are drastically better problem solvers than those with just diplomas or nothing at all.
Naturally there are exceptions, too. I've seen people with degrees (even from the university at which I work) that shock and amaze me with their stupidity. But, like I said, over all, a degree counts. And yes, I have tossed a resume because it didn't indicate a degree.
You can have as many machines as you want running other Redhat versions, but you aren't allowed to install RHAS on more machines than you have service contracts for.
I'm not allowed? Assuming on the other servers for which I have no service agreement I install only the Free software packages from RHAS, Red Hat can't tell me I'm not allowed to do that.
Red Hat certainly can refuse to provide support for those other servers. But that's all they can do.
Your interpretation does sound right. Their license agreement says you need to buy additional services from Red Hat if you want to install RHAS on other servers. But that's not accurate. You need to buy additional services from Red Hat if you want Red Hat to support that install. (And assuming on the other servers you install only Free software.)
I've always been confused by this section in Red Hat's license. I'd really like to see Red Hat clarify it. The letter of it might be unambiguous, but enough people are being confused by it that it deserves to be elaborated on.
Sure, one could use awk or perl to count lines in files, too. One could use awk or perl to recursively search for a glob. One could use awk or perl to summate a series of numbers. (And you'll notice my original post has perl code that is virtually identical to your awk code.) But The Unix Way is to have small tools that do a single thing well.
You could make your own summate script (and I have, since I find I often need that sort of tool), but it's not portable unless it's in (text|core)utils. Something like that should be, IMO.
Unix needs a tool that takes numbers via stdin (one per line) and outputs the total to stdout. I'd call it summate(1). Does anyone know of one? You could then use that to replace that ugly perl line.:)
Slashdot Mirror
My understanding is that 2.6 improves responsiveness (latency) and, quite inevitably, sacrifices raw throughput. Application load time is all about throughput, so it's probably not going to be the case that dropping in a 2.6 kernel makes applications start faster. It seems to me that this will need to be addressed at higher layers.
Jason.
You might not know this, but Mozilla doesn't just run on Unices. It also runs on Windows, Mac OS, and god knows what else. Most of what Mozilla does is not the way of Unix, mainly for the sake of being cross-platform.
Anyway, if you're interested in "small, sharp tools" or "one job, one program," you should look at Firebird and Thunderbird. You might be interested in knowing that this is the direction Mozilla is heading. So the Moz dev team would appear to agree with you there -- don't hold your breath about losing the disk cache, though.
Jason.
b)Many Linux contributors were originally UNIX developers
It's certainly misleading, but it's not a contradiction.
He didn't say Linux was built entirely by unrelated and unknown developers. He said numerous. Therefore this allows for many Linux contributors to be UNIX developers.
Of course, the letter certainly is riddled with errors and outright lies, but this particular one is not a (logical) contradiction.
Jason.
Care to point out where? Because I called Red Hat sales and was told to keep checking the website over the next couple of weeks as they would have some sort of formal announcement there.
Indeed, all the questions about educational discounts were seriously dodged in this interview. There were 3 separate questions posted and basically a one-liner by Szulik saying "Hey $50k sounds good, give me a call and we'll take your money" without any regard to the real questions.
Jason.
I have noticed, during all this Verisign SiteFinder nonsense, that many people who should know better don't seem to have any idea what a 404 error is. (Not referring to the parent poster, of course, but many others.)
404 isn't what you get when you type in a non-existent domain. That depends on your browser. Error 404 is an HTTP status code that a real live webserver returns when you request a document it can't find.
The only way you're going to get a 404 error is if you type in a VALID domain, and so 404's have nothing to do with Site Finder at all.
</rant>
Jason.
I never much liked Verisign in the past, but since I already had an account there, using them to register new domains was simply the path of least resistance. But their SiteFinder is the straw that broke the camel's back.
Jason.
I have encoded and compressed my entire DVD collection into AVI (mpeg4). I don't compress the audio, so of course I get the same AC3 stream I get when I watch from the DVD, and with constant quality encoding, the video quality is such that I usually can't tell the difference. (Of course mpeg4 has its weak points, like dark scenes, or foggy/smoky scenes.)
The disavantage of this is that I don't get DVD menus and special features, but if the movie is all you're interested in, you can squeeze it down into about 1.5G (give or take) and your common movie enthusiast (of which I am one, with a home theatre and all) would find perfectly acceptable in quality.
Jason.
Adding a router does not extend the segment. It creates a new segment and a new subnet. The 5-4-3 rule does not apply to routers. Just imagine how broken the Internet would be if we could have at most 4 routers between end points. :)
Jason.
Well, I've increased the upstream bandwidth allowed for that box, and upped the max number of http processes. *fingers crossed*
But didn't you just say:
A janitor sticking his mail password on his monitor might not be so disastrous, but, as you say, stepping stone ...
If people want to write their passwords in their wallet (with no reference to which account it is), because, well, to borrow Schneier's (paraphrased) words, "I have a life time of experience keeping my wallet safe." But post-it notes by the keyboard is definitely where I draw the line -- even if it is only a janitor.
Jason.
That's really too bad. The divx player at divx.com was the only decent player I've seen that manages to properly play video created by mencoder.
Does anyone know of other options?
Jason.
I'm a very happy Evo user, but I did check out Columba just to see what it was all about. It's a bit rough around the edges, but it was actually very functional, and I was impressed at its speed, considering it's Java.
Jason.
In related news, Microsoft, also well-known for their Rolling Stones Start Me Up campaign in '95, have signed a deal with the Olsen Twins and will begin filming their new movie, "Passport to Microsoft," expected to be released summer of 2004.
At the risk of being redundant, I have to emphatically agree with this. A few years ago I started a project that required me to wrap a C library as a python module. (The project was ORBit-Python.) Having done a lot in perlXS before that, I was quite prepared to struggle with the Python/C API.
But it wound up being truly a breath of fresh air. There are a few sticky points to get hung up on, like what functions return borrowed or new references, but the syntax is elegant and consistent, and the Python code itself is completely intuitive and a pleasure to read.
Jason.
My head just exploded trying to figure that out. Damnit, what a mess.
I haven't lost anything. No such point was made to me.
Does your *realtime* MPEG4 encoder do realtime, without noticible artifacts? I think not.
I wasn't aware there is an option for lossless compression specified in MPEG4. If that's the case (and I don't believe it is, but I could be wrong), then you're certainly right. But assuming MPEG4 is, as I believe it to be, strictly a lossy codec, then you're not going to get around artifacting in your video without applying smoothening filters that would only further separate the compressed video from its original. That's the sacrifice you make with lossy codecs.
I don't care to argue about processor specifics. It's completely irrelevant. I simply posted to refute the implied claim of parent poster that x86 is incapable of doing MPEG4 in real-time.
I use mencoder (mplayer) to do my encoding. I use CBR because it's slightly less CPU intensive. CBR is, by its very nature, more likely to produce compression artifacting. I can do VBR (constant quantizing) in real-time, but I prefer to have the wiggle room in CPU usage, and since the video source is television, I don't notice the quality difference. But I have not seen MPEG4 at sane bitrates (less than 2500kbit, say) that did not yield perceptible artifacting. I'd certainly be interested in seeing your examples.
Cheers,
Jason.
My x86 PC does that trivially. It's a 2-year-old Athlon 1400. I record TV realtime in mpeg4 (2500kbit) and mp3 (160kbit) with 30-40% CPU to spare.
Jason.
At least in Canada, there are distinct differences between the education you get behind diplomas and degrees. Depending on the position for which we are hiring, we may prefer someone with a stronger academic and theoeretical background, because it's been my experience that someone with a good education in theory picks up (and more importantly, understands) practices a lot better than someone with a diploma, and hence a solid practical knowledge, picks up the theory.
It's easy enough to go through the motions for installing software, or adding a user, or tossing in an ACL on a Cisco router. But when it comes time to actually design software or design a network, I'll tend to prefer someone with a degree who has taken courses in software engineering and computational theory.
My understanding is that things are a little different in the US, if that's where you happen to live. I believe there are fewer strict requirements on what a school is able to call a degree or a diploma, whereas in Canada those things are a little more clearly defined.
Jason.
But when it comes down to roughly equivalent work experience, a degree (not a diploma, I don't care about diplomas) will tip the balance, even if the applicant without a degree might look slightly better on paper. When it comes down to it, you have to pick a short list, and whether or not the applicant has a degree is a good metric to use.
Of course I understand what you're saying. Tossing a resume of someone with 40 years of experience but no degree when he's up against a recent graduate is ludicrous. Still, by and large, it's been my experience that people with degrees (which, when you're reading a resume, is about as good an indication as you can get about someone's academic and theoretical background) are drastically better problem solvers than those with just diplomas or nothing at all.
Naturally there are exceptions, too. I've seen people with degrees (even from the university at which I work) that shock and amaze me with their stupidity. But, like I said, over all, a degree counts. And yes, I have tossed a resume because it didn't indicate a degree.
Jason.
I'm not allowed? Assuming on the other servers for which I have no service agreement I install only the Free software packages from RHAS, Red Hat can't tell me I'm not allowed to do that.
Red Hat certainly can refuse to provide support for those other servers. But that's all they can do.
Your interpretation does sound right. Their license agreement says you need to buy additional services from Red Hat if you want to install RHAS on other servers. But that's not accurate. You need to buy additional services from Red Hat if you want Red Hat to support that install. (And assuming on the other servers you install only Free software.)
I've always been confused by this section in Red Hat's license. I'd really like to see Red Hat clarify it. The letter of it might be unambiguous, but enough people are being confused by it that it deserves to be elaborated on.
Jason.
You could make your own summate script (and I have, since I find I often need that sort of tool), but it's not portable unless it's in (text|core)utils. Something like that should be, IMO.
Jason.
[tack@draco linux]$ find -name '*.[chS]' | xargs wc -l | grep total | perl -nae '$sum += $F[0]; END { print $sum }'
4651160
Unix needs a tool that takes numbers via stdin (one per line) and outputs the total to stdout. I'd call it summate(1). Does anyone know of one? You could then use that to replace that ugly perl line. :)
Jason.
Jason.
What about the poor MIAA or RPAA that you're substing into non-existance!?
Jason.
Sure it's not anamorphic DVD, but it's certainly a lot better than 144 lines. :)
Jason.