This article argues that next-generation programming systems can accomplish this by combining three specific technologies:
-> Compilers, linkers, debuggers, and other tools that are frameworks for plug-ins, rather than monolithic applications.
-> Programming languages that allow programmers to extend their syntax.
-> Programs that are stored as XML documents, so programmers can represent and process data and meta-data uniformly.
1 - Compilers with plug-in architectures - GCC anyone? I know, he probably means something quicker and easier than writing new front- and back- ends for the Gnu Compiler Collection, but the concept is already out there.
2 - Just about any modern language does this to some degree depending on your definition. Under even the most rigorous definition of this, the good old language LISP does it with flair. Users can extend LISP syntax with ease, and user-added extended LISP syntax is virtually indistinguishable in style and functionality from the built-in elements of the language.
3 - Since existing languages have a well-known syntax which is easily machine parseable (in fact, that's what the parser and compiler do every time you use them on your source code), existing computer languages are already in a format which allows easy conversion into other formats and representation, and the gathering of metadata. Converting semicolons, whitespace, and parentheses (or whatever your language of choice uses) to xml tags doesn't really change anything, except to make things uglier and harder to type.
They may as well plan their updates for the first Tuesday of each month, as Microsoft will be incapable of responding to the new threat in time to roll the fix into the upcoming patch a week away anyways.
And also that the traditional unix system calls have been re-implemented, patched, tweaked, and re-implemented all over again by numerous vendors. The code under the hood in officially licensed unix implementations isn't always much better than linux in terms of maturity and widespread testing.
Decide right off the bat if you intend to be a Graphics Designer who specializes in doing web layouts, or a Web Coder. The two get confused and munged together, but most people are only good at one or the other. If you're going graphics design, hook up with someone who can do the coding side, and you do the HTML templates and the images and the layout and css and all that junk. If you're going the code route, learn the obvious back-end scripting languages, and learn javascript and (d)html stuff. Don't ever forget on the coder side to always think through what's happening at the very lowest levels - what bytes are being sent where and why. Don't get lost in the higher level languages, or you quickly lose sight of real efficiency and security concerns.
Generally when the term "crosstalk" is used in reference to ethernet and similar technologies, what they are referring to is basically the TX pair and RX pair in one cable interfering with each other, as opposed to two whole seperate cables interfering.
And you're right that cable quality can affect latency, but only indirectly. The real latency of the packets is unaffected, but if your bad cabling causes lots of transmission errors, then the packets have to be re-transmitted more often, causing latency up at the higher layers of the stack.
Another tip - where copper signal cabling (ethernet, serial, scsi, whatever it may be) has to come near power cabling, always cross them at right angles instead of running them parallel to each other, this greatly reduces the chances of inductive interference.
Don't forget that 10G Fiberchannel will probably come at some point as well. And in the much more realistic and cool world, there's also 2.5Gb and 10Gb Infiniband. If the Infiniband guys will just re-engineer the god-awful physical connectors they designed, it could really take hold fast. Plus it was designed as a generic high speed low-latency data transport - you can do storage, IP networking, direct MPI library stuff, etc all through one Infiniband connection. Some of the Infiniband switch vendors are also making products that adapt legacy networks onto Infiniband - you stick GigE and FC cards in the switch itself, plug them into the legacy IP and Storage networks, and they become shared virtual adapters for all the Infiniband hosts on your Infiniband network.
I agree as well. I think Planetside, for all of it's horrible faults, was a really good eye-opener on this level. Planetside was basically an "MMOFPS", but in a new and interesting way it showed, as plain as day, that MMO-[FPS/RPG] was really all one big category, and that there was a lot of room left for designing an MMORPG without the traditional timesinks.
According to your own link, they already did W in 2000. He has now joined a very exclusive list of people to make TMotY twice.
Re:Examples of Math books for lay people
on
Prime Obsession
·
· Score: 2, Insightful
I second this - Mathematics for the Million is truly a classic that belongs in this category that the story author referred to. It should be handed out to every child at a reasonable age, so that they can use it as an extra-curricular guide as they learn their way through the horrors of school-taught math.
There is really only one good reason to ever use RAID5, and that is that you're too tight on money to be able to afford to RAID1 (Mirror) the storage you need (If you need 400G of space, RAID1 is gonna cost you 800G of storage, whereas RAID5 might only cost you 500G of storage). RAID1 is both faster (For writes and especially reads) and more resilient than RAID5. Assuming you can afford it (and storage itself is pretty cheap today, especially if you don't get a fancy RAID5 controller), just go with RAID1.
If you want really nice performance and you're buying 4+ drives, do RAID1+0 - mirror the drives up in pairs (where the pairs are as diverse as your setup allows, seperate controllers and/or chassis and/or power, etc...), then stripe the data volume on top of the sets of mirror-pairs.
The World Calendar fails at some thigns it specifically is meant to address. They speak of making statistics and whatnot easier by having even, regular, periodic numbers for days, weeks, months, years, etc. However, the neccesity of the two "W" holidays which they ignore greatly screw this up. They act like when a W day happens, the entire world simply stops and just stares at the wall for 24 hours, and afterwards we forget that the W-day ever existed, which is simply not true. My weather-data-monitoring station will still be taking temperature measurements all through W, what date does it log them under? And once I've got my whoel year of temperature data, do I simply throw out my W data, or do I go with imperfect analysis, since W made some months/quarters/years longer than others? etc....
It might be more accurate than it was 50 or 60 years ago, but then again my powers of ESP have grown exponentially in the last 10 years too. I used to have a 0.001% success rate, now I'm up to 1%.
Seriously, I get weather from all over - I look at raw NOAA data, I look at weather.com, I see the local forecasters with their Doppler Radar crap, I get weather messages every morning on my cellphone even. They're all hidieously inaccurate. I really *do* have a better idea how the weather will be on any given day if I ignore everything else and watch the barometer on my balcony, than if I read the forecasts. It's true.
This guy hasn't a prayer of getting his calendar implemented. He's a nutcase, and his calendar is riddled with practical problems (which he even notes on his site amongst the "FAQs", and then brushes aside with illogical retorts). As further proof of his unfitness as an architect of serious systems for human use, in another part of his calendar site, he gives code examples in Fortran. Anyone who, when given the chance to write a code example in order to explain a simple calendar concept, immediately goes for Fortran as his language of choice, is not someone I want designing anything that might affect my life.
I'd like to see how they plan on monitoring my mage as it talks to your cleric in some obscure, nearly impossible to reach (unless you're level 50) corner of our favorite MUD.
With a sniffer, a tiny fragment of code which decodes the protocol used by your game, and perhaps some decryption routines to get around the probably ill-designed obfuscation the game designers put in place, the same way they do everything else. It's not hard. It's the same way you sniff any other traffic. Do you really think the FBI has thousands of McJobbers actually running irc clients and manually creating account in web forums and reading posts? They just sniff the traffic, it's all IP. Only red flags from the traffic logs warrant further manual investigation that might involve logging into the discussion medium of choice in a traditional fashion.
Isn't the current weather prediction pretty much like the 1800's anyways? I get a better predictability just by watching the barometer on my patio that by looking at the official forecasts.
... Or lack thereof in the case of some various linux and other *nix boot processess. The first step in booting faster is to actually know for sure which things depend on which other things. The ancient simplistic approach was linear ordering (The "runlevels" are a higher-order linear ordering, and then within each runlevel the services were also number sequentially). If a linear ordered list is all you've got, then you're gauranteed to be doing a worst-case serial execution of all the startup tasks, which sucks.
Have a look at Gentoo's init script setup. It actually knows about hard and soft dependencies. It still only fires off one script at a time at the moment, but importantly the system has all the right data in hand to parallelize the process. (Hard dependencies are for instance when then nfsclient script *requires* that the network script is run before it, soft dependencies are things like the apache script saying that *if* the mysql service is enabled, please start it before you start me, but I do not require it if it wasn't enabled explicitly by the admin).
Gentoo would be a great jumping point for parallel ization of startup tasks. The only real issue is screen clutter.... but I think that can be solved (if not terribly elegantly) by line-buffering the startup messages and displaying them serially in whatever order they "complete", resulting in a random line-order on the screen, but nothing stepping on each other's lines.
I agree. One webapp I wrote a while back I ended up needing to put the SHA1 hash algorithm on the client-side in javascript and I did the same thing in order to keep things as snappy as possible. I ripped it down to the smallest possible size (by hand) by renaming all the variables and function names as short as possible and eliminating all useless whitespace, etc. As a side effect the source became fairly obfuscated like Google's looks, not that I cared to obfuscate sha1.
Linux's long-term future has always been a bit of a question mark. Who's to say that 10 years from now something new in the open-source world might not emerge and overtake it, taking many things from it? It's always possible, and it seems that _eventually_ it would inevitably happen as the mood of the open-source community will eventually shift to feeling that linux is crufty and a fresh start is needed.
But all of that being said, I think I can say two things with a fair amount of certainty:
1) That time is not now. Linux is really just getting into it's game. It has lots of growth and evolution left in the commercial world. The time may come, but not in the next 4-5 years for damn sure. Even then, while the industry may begin to swing away from Linux to something new, Linux will remain a strong force for many many years to come during the transition.
2) Whatever that New Thing is that comes along to supplant Linux as the new Open Source Darling Operating System, it damn sure won't be Solaris.
I'm a coder-admin too. The title changed from time to time over the past 10 years or so as I've moved from job to job. "Senior Systems Engineer", "Unix Systems Administrator", "Network Administrator", whatever. In any case, my job has always been to do all the "systems" and "network" stuff, and it's very much a Systems Admin role more than a Coder role. But at the same time, I think my primary strength in these roles is that I love to code and do it well. I feel as a systems guy, my primary and asymptotic goal is to put myself out of a job - to code and automate everything related to what I do as much as possible in the hopes of my job becoming unneccesary if I did it well enough. That can't really happen, but it's just a trajectory to put yourself on, if you know what I mean.
And in contrast to the admin who posted earlier, I really dig coding. My best weeks at work are the weeks where I get to code uninterrupted for the whole week. This usually involves pawning off work on other sys-admins in the group who don't or won't code. But I would never consider myself a "coder", because I'll probably never work on a commercial software product. My programming is "systems" programming, if you get my drift.
Check out PopCap games. I heard about them because one of the audio guys from the old early 90's asm demo crew Future Crew ended up working there. They make kick-ass simple games with no adware or spyware. You can play limited versions of the game in flash in your browser, or buy cheap licenses and download them for Wintel. www.popcap.com.
I've been in the same boat as the guy you're replying to, and my answer to this question would have been that what I really wanted was something structured and designed more like the typical CMS implementation (database-backed, web-based admin without any html coding experience needed on the users' part, "document" upload of word/pdf/etc with searches and categories and all that, etc...), but I just don't want "community" features like blogging, news, rss, etc...
The usual answer that I've taken is to use one of the full-blown CMS-like packages out there and strip out all the functionality I don't want, which can be a pain to maintain as new releases come out.
IHNRTFA (I have not read the f-in article), but it seems to me this is yet another case of the misuse of statistics. Just because X% of PopulationY happens to also be in PopulationZ doesn't imply a causal relationship in any specific direction. It's just as likely that the real reason for the link is the opposite of what they suppose: that persons who have the genetic defects and/or environmental factors in their lives which lead to a higher rate of glaucoma are more likely to be computer users....
If you're looking for performance, ditch RAID5 and go with RAID0+1 (mirror pairs of disks, then stripe the mirrors). It costs more in terms of dollars per gigabyte, but the performance and reliability increases are substantial over RAID5. The only justification for RAID5, really, is "we can't afford to buy enough disks to do something else" (which is a valid argument for many people). The other thigns you touched on still apply in a stripe/mirror world as far as splitting things over controllers and whatnot.
Obviously different people call fall all over that two-dimensional map, and thus there are such things as left-libertarian and right-libertarian -ish people. In counterpoint to the lefty above, I'm a republican-leaning libertarian. This is really more my gaming the system than anything though. I feel that if the democrats succeed, the ways in which they deviate from libertarianism (primarly thinking about social programs and medical care issues) will increasingly become solidified in US law, whereas I feel the major way the republicans lean away from libertarianism is their moral/religious laws, which I feel are inevitably going out the window in the near future no matter what due to the shifting of society away from hardcore religion.
Slashdot Mirror
1 - Compilers with plug-in architectures - GCC anyone? I know, he probably means something quicker and easier than writing new front- and back- ends for the Gnu Compiler Collection, but the concept is already out there.
2 - Just about any modern language does this to some degree depending on your definition. Under even the most rigorous definition of this, the good old language LISP does it with flair. Users can extend LISP syntax with ease, and user-added extended LISP syntax is virtually indistinguishable in style and functionality from the built-in elements of the language.
3 - Since existing languages have a well-known syntax which is easily machine parseable (in fact, that's what the parser and compiler do every time you use them on your source code), existing computer languages are already in a format which allows easy conversion into other formats and representation, and the gathering of metadata. Converting semicolons, whitespace, and parentheses (or whatever your language of choice uses) to xml tags doesn't really change anything, except to make things uglier and harder to type.
They may as well plan their updates for the first Tuesday of each month, as Microsoft will be incapable of responding to the new threat in time to roll the fix into the upcoming patch a week away anyways.
And also that the traditional unix system calls have been re-implemented, patched, tweaked, and re-implemented all over again by numerous vendors. The code under the hood in officially licensed unix implementations isn't always much better than linux in terms of maturity and widespread testing.
Decide right off the bat if you intend to be a Graphics Designer who specializes in doing web layouts, or a Web Coder. The two get confused and munged together, but most people are only good at one or the other. If you're going graphics design, hook up with someone who can do the coding side, and you do the HTML templates and the images and the layout and css and all that junk. If you're going the code route, learn the obvious back-end scripting languages, and learn javascript and (d)html stuff. Don't ever forget on the coder side to always think through what's happening at the very lowest levels - what bytes are being sent where and why. Don't get lost in the higher level languages, or you quickly lose sight of real efficiency and security concerns.
Generally when the term "crosstalk" is used in reference to ethernet and similar technologies, what they are referring to is basically the TX pair and RX pair in one cable interfering with each other, as opposed to two whole seperate cables interfering.
And you're right that cable quality can affect latency, but only indirectly. The real latency of the packets is unaffected, but if your bad cabling causes lots of transmission errors, then the packets have to be re-transmitted more often, causing latency up at the higher layers of the stack.
Another tip - where copper signal cabling (ethernet, serial, scsi, whatever it may be) has to come near power cabling, always cross them at right angles instead of running them parallel to each other, this greatly reduces the chances of inductive interference.
Don't forget that 10G Fiberchannel will probably come at some point as well. And in the much more realistic and cool world, there's also 2.5Gb and 10Gb Infiniband. If the Infiniband guys will just re-engineer the god-awful physical connectors they designed, it could really take hold fast. Plus it was designed as a generic high speed low-latency data transport - you can do storage, IP networking, direct MPI library stuff, etc all through one Infiniband connection. Some of the Infiniband switch vendors are also making products that adapt legacy networks onto Infiniband - you stick GigE and FC cards in the switch itself, plug them into the legacy IP and Storage networks, and they become shared virtual adapters for all the Infiniband hosts on your Infiniband network.
I agree as well. I think Planetside, for all of it's horrible faults, was a really good eye-opener on this level. Planetside was basically an "MMOFPS", but in a new and interesting way it showed, as plain as day, that MMO-[FPS/RPG] was really all one big category, and that there was a lot of room left for designing an MMORPG without the traditional timesinks.
According to your own link, they already did W in 2000. He has now joined a very exclusive list of people to make TMotY twice.
I second this - Mathematics for the Million is truly a classic that belongs in this category that the story author referred to. It should be handed out to every child at a reasonable age, so that they can use it as an extra-curricular guide as they learn their way through the horrors of school-taught math.
There is really only one good reason to ever use RAID5, and that is that you're too tight on money to be able to afford to RAID1 (Mirror) the storage you need (If you need 400G of space, RAID1 is gonna cost you 800G of storage, whereas RAID5 might only cost you 500G of storage). RAID1 is both faster (For writes and especially reads) and more resilient than RAID5. Assuming you can afford it (and storage itself is pretty cheap today, especially if you don't get a fancy RAID5 controller), just go with RAID1.
If you want really nice performance and you're buying 4+ drives, do RAID1+0 - mirror the drives up in pairs (where the pairs are as diverse as your setup allows, seperate controllers and/or chassis and/or power, etc...), then stripe the data volume on top of the sets of mirror-pairs.
The World Calendar fails at some thigns it specifically is meant to address. They speak of making statistics and whatnot easier by having even, regular, periodic numbers for days, weeks, months, years, etc. However, the neccesity of the two "W" holidays which they ignore greatly screw this up. They act like when a W day happens, the entire world simply stops and just stares at the wall for 24 hours, and afterwards we forget that the W-day ever existed, which is simply not true. My weather-data-monitoring station will still be taking temperature measurements all through W, what date does it log them under? And once I've got my whoel year of temperature data, do I simply throw out my W data, or do I go with imperfect analysis, since W made some months/quarters/years longer than others? etc....
It might be more accurate than it was 50 or 60 years ago, but then again my powers of ESP have grown exponentially in the last 10 years too. I used to have a 0.001% success rate, now I'm up to 1%.
Seriously, I get weather from all over - I look at raw NOAA data, I look at weather.com, I see the local forecasters with their Doppler Radar crap, I get weather messages every morning on my cellphone even. They're all hidieously inaccurate. I really *do* have a better idea how the weather will be on any given day if I ignore everything else and watch the barometer on my balcony, than if I read the forecasts. It's true.
This guy hasn't a prayer of getting his calendar implemented. He's a nutcase, and his calendar is riddled with practical problems (which he even notes on his site amongst the "FAQs", and then brushes aside with illogical retorts). As further proof of his unfitness as an architect of serious systems for human use, in another part of his calendar site, he gives code examples in Fortran. Anyone who, when given the chance to write a code example in order to explain a simple calendar concept, immediately goes for Fortran as his language of choice, is not someone I want designing anything that might affect my life.
With a sniffer, a tiny fragment of code which decodes the protocol used by your game, and perhaps some decryption routines to get around the probably ill-designed obfuscation the game designers put in place, the same way they do everything else. It's not hard. It's the same way you sniff any other traffic. Do you really think the FBI has thousands of McJobbers actually running irc clients and manually creating account in web forums and reading posts? They just sniff the traffic, it's all IP. Only red flags from the traffic logs warrant further manual investigation that might involve logging into the discussion medium of choice in a traditional fashion.
Isn't the current weather prediction pretty much like the 1800's anyways? I get a better predictability just by watching the barometer on my patio that by looking at the official forecasts.
... Or lack thereof in the case of some various linux and other *nix boot processess. The first step in booting faster is to actually know for sure which things depend on which other things. The ancient simplistic approach was linear ordering (The "runlevels" are a higher-order linear ordering, and then within each runlevel the services were also number sequentially). If a linear ordered list is all you've got, then you're gauranteed to be doing a worst-case serial execution of all the startup tasks, which sucks.
Have a look at Gentoo's init script setup. It actually knows about hard and soft dependencies. It still only fires off one script at a time at the moment, but importantly the system has all the right data in hand to parallelize the process. (Hard dependencies are for instance when then nfsclient script *requires* that the network script is run before it, soft dependencies are things like the apache script saying that *if* the mysql service is enabled, please start it before you start me, but I do not require it if it wasn't enabled explicitly by the admin).
Gentoo would be a great jumping point for parallel ization of startup tasks. The only real issue is screen clutter.... but I think that can be solved (if not terribly elegantly) by line-buffering the startup messages and displaying them serially in whatever order they "complete", resulting in a random line-order on the screen, but nothing stepping on each other's lines.
I agree. One webapp I wrote a while back I ended up needing to put the SHA1 hash algorithm on the client-side in javascript and I did the same thing in order to keep things as snappy as possible. I ripped it down to the smallest possible size (by hand) by renaming all the variables and function names as short as possible and eliminating all useless whitespace, etc. As a side effect the source became fairly obfuscated like Google's looks, not that I cared to obfuscate sha1.
Linux's long-term future has always been a bit of a question mark. Who's to say that 10 years from now something new in the open-source world might not emerge and overtake it, taking many things from it? It's always possible, and it seems that _eventually_ it would inevitably happen as the mood of the open-source community will eventually shift to feeling that linux is crufty and a fresh start is needed.
But all of that being said, I think I can say two things with a fair amount of certainty:
1) That time is not now. Linux is really just getting into it's game. It has lots of growth and evolution left in the commercial world. The time may come, but not in the next 4-5 years for damn sure. Even then, while the industry may begin to swing away from Linux to something new, Linux will remain a strong force for many many years to come during the transition.
2) Whatever that New Thing is that comes along to supplant Linux as the new Open Source Darling Operating System, it damn sure won't be Solaris.
I'm a coder-admin too. The title changed from time to time over the past 10 years or so as I've moved from job to job. "Senior Systems Engineer", "Unix Systems Administrator", "Network Administrator", whatever. In any case, my job has always been to do all the "systems" and "network" stuff, and it's very much a Systems Admin role more than a Coder role. But at the same time, I think my primary strength in these roles is that I love to code and do it well. I feel as a systems guy, my primary and asymptotic goal is to put myself out of a job - to code and automate everything related to what I do as much as possible in the hopes of my job becoming unneccesary if I did it well enough. That can't really happen, but it's just a trajectory to put yourself on, if you know what I mean.
And in contrast to the admin who posted earlier, I really dig coding. My best weeks at work are the weeks where I get to code uninterrupted for the whole week. This usually involves pawning off work on other sys-admins in the group who don't or won't code. But I would never consider myself a "coder", because I'll probably never work on a commercial software product. My programming is "systems" programming, if you get my drift.
Check out PopCap games. I heard about them because one of the audio guys from the old early 90's asm demo crew Future Crew ended up working there. They make kick-ass simple games with no adware or spyware. You can play limited versions of the game in flash in your browser, or buy cheap licenses and download them for Wintel. www.popcap.com.
I've been in the same boat as the guy you're replying to, and my answer to this question would have been that what I really wanted was something structured and designed more like the typical CMS implementation (database-backed, web-based admin without any html coding experience needed on the users' part, "document" upload of word/pdf/etc with searches and categories and all that, etc...), but I just don't want "community" features like blogging, news, rss, etc...
The usual answer that I've taken is to use one of the full-blown CMS-like packages out there and strip out all the functionality I don't want, which can be a pain to maintain as new releases come out.
IHNRTFA (I have not read the f-in article), but it seems to me this is yet another case of the misuse of statistics. Just because X% of PopulationY happens to also be in PopulationZ doesn't imply a causal relationship in any specific direction. It's just as likely that the real reason for the link is the opposite of what they suppose: that persons who have the genetic defects and/or environmental factors in their lives which lead to a higher rate of glaucoma are more likely to be computer users....
If you're looking for performance, ditch RAID5 and go with RAID0+1 (mirror pairs of disks, then stripe the mirrors). It costs more in terms of dollars per gigabyte, but the performance and reliability increases are substantial over RAID5. The only justification for RAID5, really, is "we can't afford to buy enough disks to do something else" (which is a valid argument for many people). The other thigns you touched on still apply in a stripe/mirror world as far as splitting things over controllers and whatnot.
Obviously different people call fall all over that two-dimensional map, and thus there are such things as left-libertarian and right-libertarian -ish people. In counterpoint to the lefty above, I'm a republican-leaning libertarian. This is really more my gaming the system than anything though. I feel that if the democrats succeed, the ways in which they deviate from libertarianism (primarly thinking about social programs and medical care issues) will increasingly become solidified in US law, whereas I feel the major way the republicans lean away from libertarianism is their moral/religious laws, which I feel are inevitably going out the window in the near future no matter what due to the shifting of society away from hardcore religion.