You're right, sorry. Her husband was, I believe, Lord Something or Other which entailed some sort of geographical or conceptual region to which Lovelace is appended. Ada having had about five names during her life does not help in the slightest with this.
There seems also to have been much dispute between her, her mother, and her youngest son, from which the gambling/druggie description may have been produced. Some dispute on that point.
Regardless of this, I think Mr. Glass is a first class twit-of-the-media and should be debunked as often as possible, and as publicly as possible.
One of the leading occupations of media twits is debunking other media twits. It makes for long, self-righteous columns unravelling other long, self-righteous columns. That gets added to a simpering "hard-news" corps whose main function is to give any new product its alottment of drool and "Can it beat X?"-type "analysis" pieces. Then add a lot of pandering to the big-name advertisers, and you have... the US technology press, both print and electronic.
(and, of course, I'm saying this while reading slashdot. oh well.:))
Like much of that article, that bit seemed to be a mixture of journalistic cynicism, journalistic naivite and journalistic arrogance.
I wasn't able to decide if the author was trying to make jabs at the OSS realm or not -- he dismissed the GPL aspect of BO2k with the "obfuscation" claim, missed every ramification of an open source BO except for the concern of the script kiddies about trojaned exploits.
(aside: Kiddies don't read source. The claim that BO might be obfuscated in the identifier/whitespace sense is bogus -- it would reduce the point of GPLness to a PR tactic which would be quickly noted and cDc would be reviled for it, more than they already are. Obfuscation in the code-structure sense would merely make it unmaintainable, not unusable or unmodifiable)
... and, to resume, he seemed generally to propose (especially with your quoted excerpt) that the darker side of security research is somehow wrong and misguided and should go away (gosh, someone should tell that to street hoodlums), and that open-spec/open-source/open-attack security is somehow a bad thing. He did get right the part about how there's no common code of ethics -- an attribute he might find is shared by many sectors of street criminals, marketing executives and politicians.
He mentions also that defcon's a party, which is true enough, but then forgets that fact for the rest while applying his lofty judgement to the various frivoloties. Defcon is supposed to be gross, overstated and stupid -- it's a party. It's not a particularly serious meeting of minds, in any sense, and interpreting it as such leads to all sorts of depressingly absurd conclusions, such as those found in this article.
True enough, though in the event of some sort of backbone catastrophe, routers configured to silently deny traffic on port 80 in favor of ports 21-25 would raise the survival factor immensely. The fault-tolerance algorithms are still there, but web (and to a lesser extent mail) usage has eliminated hope for redundant carrier channels for the time being.
(offtopic) Rather like some of the old Atari (or it might have been Amiga) demos that ran the main code loop on the CPU, and used the tiny processor in the keyboard to do rasterization, because, well, it was programmable.:)
(more ontopic) This is nice to see, though -- any modern remake of lpr/lpd would be an improvement, and in particular seems a good example of the potential for coexistence between commercial for-profit software and free software -- the architecture is free, and specialized drivers for individual printers can be had for cost. Might not be suitable for home and hackerish uses in that respect, but business environments would lap it up.
Beg pardon, but Ada Byron/Lovelace's code was written for Babbage's Analytical Engine. Her slacker husband was the poet Lord Byron. Ada married Byron, and did her work with Babbage on the engines. Babbage != Byron; in fact they were some distance apart, though it seems that Byron was fairly supportive of Ada's work, even though at the time the involvement of women in science and mathematics was fairly limited and to some degree discouraged. Ada was herself encouraged by her (cretin) mother to study math, because she (Ada) was getting interested in boys, and her mother (who hated men in general) feared for her virtue. Turned out that Ada liked numbers as much as boys, so her mother did succeed in turning her off men, or at least men who didn't crunch numbers, until her (arranged) marriage to Byron. Ada had three kids and raised them all, whilst working with Babbage.
Babbage himself, not having the advantage of any formal procedures of computer design, never really finished either Engine, since he got distracted by the Difference Engine by the prospects of the Analytical Engine, and the latter was largely doomed by the inadequacy both of mechanical calculators and of machining techniques in England and the world at large of the period (roughly 1830-death, since Babbage never kept up even after being discredited).
Not sure if Ada qualifies as a geek or not; the attributes of geekiness hadn't really been established, and she was required to also be Lady Ada Byron of Lovelace, which is probably like wearing a suit and such. Maybe she was just really smart and good with numbers -- there's not much record of what she herself was like, other than wrt math.
Her programs have all been published, BTW; the one for calculating the seventh Bernoulli number is used in math texts, including the bug and the design flaw.:)
FWIW, Sonoma State University, California (part of the CSU system mentioned in the essay) has one remaining NT server left in CS and IT of which I'm aware. The rest, in order of occurrence, are Linux, Solaris, MacOS and VMS. The IT dept is almost entirely UNIX people. CS is a mixed bag -- the programming classes are all taught on CodeWarrior under MacOS and NTW, but the dominant sentiment is that NT is a huge drag and not worth keeping around. Every PC in the labs has Linux as its default boot OS. There's a lot of resentment of MS and its products around the department, from both students and profs. I haven't encountered what some other posters have reported, where profs refuse to teach on NT -- but SSU is a heavily Mac-dominated campus and most of them prefer the Macs and teach on those.
I would infer that an MS strategy to try to undercut the servers would only work if MS marketed to those parts of an academic structure which don't know anything about computers -- which is to say, the administrators, deans, etc., who make the decisions and spend the money and don't actually teach or do much of anything.
Also at issue, of course, is money -- budgets for CSU have been getting cheerfully cut by the CA legislature for some time now, and CS has had its time-to-graduation increase by 25% or so because of restricted class availability (this works out to be more profitable for the institution, which gets paid by enrollment by the gov't plus tuition per student per semester). With some flag-waving, Linux can win out in that respect -- "discounted" can't compete with "free," though the problem is largely the same as competing for prestige with corporate purchasing execs who know little beyond what's written on advertisements.
It's a wholly valid line of discussion. Commercial software has gender-targeting; Linux users get most of their software through their distribution. Thus it's a useful thing to reason through. I don't think it's an especially good idea, but it is a useful thing to consider, esp. wrt the software of interest to women.
FWIW, open source software is much friendlier overall to women -- most of our software makes things and does things -- and those aren't specific to males more than females any more than hammers or screwdrivers. Most of the gender differentiation is based on games, and of our games, we've got a few gore/violence/etc games, and quite a lot of stuff with redeeming qualities like mental exertion and stuff. Not doing too bad there.
For the OSS community as a whole, we've got a long way to go before this stuff can approach gender parity -- and attitudes like yours are, sadly, part of that long way.
IME, you have the lawyer part pretty much correct. IBM has always had a large fleet of those. Probably works both ways -- historically the lawyers have been needed to avoid getting into antitrust situations (IBM is huge, remember) -- the lawyers were the reason IBM didn't buy out Eckert and Mauchly when their fledgling company (making BINAC/UNIVACs) headed straight for the ground. Watsons Sr. and Jr. prob. would have liked to, and E&M wanted to sell it and go back to doing the engineering work, but the lawyers wouldn't have it.
FWIW, the MWave modem&sound card was put in a lot of laptops -- IBM doesn't like reproducing engineering. It's pretty much the same as most cards of its type -- basically a DSP, some support circuitry and a driver. It's not even that bad an idea, since a DSP-only modem uses less power and takes up less physical space, both of which are at a premium in laptops. For a long time IBM refused to release the specs to the thing, leading to hacks to make it work in SB compatibility mode. More recently (and the data on this point is confused a bit), IBM seems to have finally recanted and provided some basic specs for talking to the thing. Dale Wick has been working on making the thing work, or at least controlling it a bit, under Linux.
I saw an external Firewire drive for the first time yesterday (in a Mac computer shop, which I don't frequent, so that's likely why). Looked like an external drive with a big price tag. I'd have preferred to see firewire survive, since it had a lot of potential for network-fabric uses. It concerns me particularly that if Apple stakes much on Firewire, they could be hurt when the Wintel machine lumbers to USB without looking back. While I don't like Apple's patent/copyright tendencies, I like them much better than the Wintel congingent's, and I like having Apple around and doing well. A Wintel-dominated industry is really, really depressing.
Those who make laws in the US are very often former attorneys, or in some cases law enforcement officials or ex-military officers. Granted lawyers are intelligent people, but those who hold legislative office are generally subject to no deep understandings of anything other than bureaucratic and other governmental process. So to put it mildly, they don't seem content with a canonical cover of a set of laws.
But, looked at another way, most legislators have placed their entire faith in their own laws, and have never learned to deal with defiance. Like Aman Hannesy (sp) said to a judge whilst being prosecuted, "Aw judge, your damn laws... the good people don't need 'em and the bad people don't obey em, so what good are they anyway." Theorize: what would happen if everyone, simultaneously, ceased obeying crypto export laws?
And I had commercially sensitive data in my email (which would be stupid on a non-POP3 server)
I hope you're not inferring that it's a good idea to pass data through a POP3 server. Not sure if you've encountered this one yet, but POP3 (and most of its kindred) send passwords and mail in the clear, the same way hotmail does. Indeed hotmail would be slightly more secure, since the passwords are likely sent in a POST form, which is mime64-encoded and thus very slightly protected against casual over-shoulder interception. Further, POP is a much more common target for interception since its use is so widespread and the format is quite standardized.
"Secure mail," inasmuch as that can be taken as anything but a contradiction in terms, involves stuff like a secure transmission client, encrypted channels all the way from sender to recipient, storage in encrypted form or on a cryptographic filesystem on a trusted, isolated server, and a secure reception client. At present hardly any such systems exist. The ones that do -- well, they don't run POP3.
Re:Resolving the Uptime Syndrome
on
Kernel 2.2.12
·
· Score: 1
As soon as Linus / Alan figgers out how to upgrade an entire kernel from source without rebooting, I'll be happy! ; )
Solaris can do it... they "recommend" going singleuser first, but it amounts basically to a hot-swap kernel. I have no clue how it's done, but I'm jealous too.
I suppose in a HA cluster, you would generally count cluster uptime rather than machine uptime -- maybe then individual kernel upgrades could be discounted so long as the cluster output was still flowing. Hmm.
Even without the merger/buyout consequences, few consumers are prepared to cope with the maneuvering performed by credit companies. The linuxfund APR is a mild (there are much worse) example -- 3.9% for four months, long enough for consumers to get into more debt than they can easily pay off. Then the "introductory" period ends, the rates more than triple, and suddenly many people are left with debts that grow faster than they can pay them back off. Thence they're in thrall to one of the worst kind of creditors one can have.
And, unfortunately, there aren't any fully effective alternatives in the American consumer money tool thing.
That was my interpretation of the posting also. We've seen examples of what happens when MS controls a programming language -- it becomes virtually unusable for any real work. Hence VB and its derivatives. Not that MS even being fully acquitted of infringement in the Java case would lead directly to MS controlling Java; more likely they'd just be free to pollute Java and try to emasculate it in favor of stuff that doesn't work cross-platform. Which would be unfortunate, because Java is a fairly neat bunch of things which haven't been done together in a while.
I bought a copy of xingenc when it was released for Linux (impulse purchase). The speed was quite good, and the speed/quality ratio was also quite ideal. A while back I did a by-ear comparison of bladeenc, xing and 8hz' encode -- bladeenc and encode were both good quality, their perceptible flaws were some slight muddling when both high and low freqs were occuring simultaneously in the music (which is a waveform encoding issue we're not going to get away from). Xing's quality was perceptibly lower, audibly distorting both extremes of the frequency range (plus chopping the ones higher than you're supposed to be able to hear). That was all at 128kbps -- at 160kbps and above, I couldn't make out any difference.
OTOH, the Xing encoder supports VBR encoding, which did seem to increase the perceived quality quite a bit. The feature is available under Linux; the files are a bit bigger with a mid-to-high quality settings, but the result sounds nice.
Somewhere there was an actual frequency-analysis report of a bunch of mp3 encoders, including all three of the above. Now I've forgotten the URL.
2+2=5 was one of the precepts used in the interrogation (reeducation, brainwashing, whatever you'd rather call it) of Winston in 1984. Not the main criteria, but one of them.
"Don't do it to me, do it to Julia!"
There are a few high school English classes that cover 1984. There are a million or so more that don't. Pity, because it's a more useful text for your average public school student than much of what gets covered there.
In any case, a server should be able to survive on zero writeable disk space. It's acceptable, if undesirable, for operations that involve writing to the disk (which does not include static HTML, that being all I saw on the test site), to fail semi-gracefully.
Under most decent OSes, writes to a disk-full file on a normal filesystem will get bufferred in RAM until free space opens up; when the RAM's full, it has to start discarding those buffers, once it's pared down the cache and forced some processes out to swap. Maybe that's involved, maybe not. The suggestion that the machine has a lot of logging turned on so they could benefit from successful crashes has merit, though -- although I'd be logging it to a different machine with a huge pile of disks that was also logging all the net traffic. Oh well. "Poor MS.":)
Pray don't confuse firewalls with security. Machines behind a firewall are only as secure (from the big bad net, assuming they're connected to it, as most firewalls are) as the TCP/IP stacks and services that answer on ports accessible through the firewall and/or its sockets. Most firewalls that I've had experience with have closed off all ports except the ones that people needed to use from the outside -- and half the time that included stuff like pop[23], imap, smtp, etc., and on which the servers answering those ports tended to be way behind on their updates because people had this sense of security lent by the firewall.
Also, firewalls don't work from people who can emit packets from inside your firewall -- and that's surprisingly easy to do, either through coercion of the firewall box's network stack, compromise of a machine behind the firewall through some open port, or simply being behind the firewall in the first place (as in many corporate environments). If a firewall is configured to permit connections to ports 22 (ssh) and 443 (SSL http), there's no particular reason why an attacker can't arrange for a root shell to answer on one of those ports, and with most network installations no one would be the wiser.
#2 has an implementation benefit as well -- as the Echelon paper highlighted, none of the major US organizations responsible for the widespread electronic interferenec with our privacy have the CPU power necessary for widespread brute-force attacks on our ciphers. Therefore, one good way to oppose the US government's invasions of our privacy is to encrypt as great a volume of our traffic as possible, thus presenting the FBI, NSA etc. with so much enciphered traffic that they can't hope to ascertain what in it was wortht he CPU time. Crypto makes for good privacy against wiretapping in its own rights, and one can help thwart those who would trespass on our privacy simply by participating in overwhelming their CPUs.
In other words, encrypt everything, whether it needed it or not. By a happy coincidence, the internet is wonderful for that.:)
After perusing through them, I find also that the message boards attached to the poll reflect a level of intelligence, respect and tact that makes slashdot look like a graduate debate seminar. Between the Theologically Impaired and the Single-Issue Media Potatoes, it's a large and gratuitous mess.
Without endorsing any specific approach, when you do your architecture design, look for opportunities to make it distributable. In general it's always cheaper to buy ten average PCs than it is to buy a single quad-xeon or Sun E10k or such -- though the job may not require that much if you do your engineering right. In the scenario you've described, look for use of multiple POP/IMAP servers sharing a common data repository (or, if it's not essential to have everyone share the same disk, divide that duty up also). Check out the high-availability Linux project for useful tools. Even if you wind up building things on one big server rather than several littler ones, leave room in your software setup so that you can distribute the load later on if need be without major reengineering.
And stay the hell away from Exchange. The degree of pain to be had from that pile of dung defies description, and the pain seems to grow exponentially rather than linearly as demand rises.
Slashdot Mirror
You're right, sorry. Her husband was, I believe, Lord Something or Other which entailed some sort of geographical or conceptual region to which Lovelace is appended. Ada having had about five names during her life does not help in the slightest with this.
There seems also to have been much dispute between her, her mother, and her youngest son, from which the gambling/druggie description may have been produced. Some dispute on that point.
Regardless of this, I think Mr. Glass is a first class twit-of-the-media and should be debunked as often as possible, and as publicly as possible.
One of the leading occupations of media twits is debunking other media twits. It makes for long, self-righteous columns unravelling other long, self-righteous columns. That gets added to a simpering "hard-news" corps whose main function is to give any new product its alottment of drool and "Can it beat X?"-type "analysis" pieces. Then add a lot of pandering to the big-name advertisers, and you have... the US technology press, both print and electronic.
(and, of course, I'm saying this while reading slashdot. oh well. :))
I wasn't able to decide if the author was trying to make jabs at the OSS realm or not -- he dismissed the GPL aspect of BO2k with the "obfuscation" claim, missed every ramification of an open source BO except for the concern of the script kiddies about trojaned exploits.
(aside: Kiddies don't read source. The claim that BO might be obfuscated in the identifier/whitespace sense is bogus -- it would reduce the point of GPLness to a PR tactic which would be quickly noted and cDc would be reviled for it, more than they already are. Obfuscation in the code-structure sense would merely make it unmaintainable, not unusable or unmodifiable)
... and, to resume, he seemed generally to propose (especially with your quoted excerpt) that the darker side of security research is somehow wrong and misguided and should go away (gosh, someone should tell that to street hoodlums), and that open-spec/open-source/open-attack security is somehow a bad thing. He did get right the part about how there's no common code of ethics -- an attribute he might find is shared by many sectors of street criminals, marketing executives and politicians.
He mentions also that defcon's a party, which is true enough, but then forgets that fact for the rest while applying his lofty judgement to the various frivoloties. Defcon is supposed to be gross, overstated and stupid -- it's a party. It's not a particularly serious meeting of minds, in any sense, and interpreting it as such leads to all sorts of depressingly absurd conclusions, such as those found in this article.
Poor boardwatch. They've gone downhill.
True enough, though in the event of some sort of backbone catastrophe, routers configured to silently deny traffic on port 80 in favor of ports 21-25 would raise the survival factor immensely. The fault-tolerance algorithms are still there, but web (and to a lesser extent mail) usage has eliminated hope for redundant carrier channels for the time being.
(more ontopic) This is nice to see, though -- any modern remake of lpr/lpd would be an improvement, and in particular seems a good example of the potential for coexistence between commercial for-profit software and free software -- the architecture is free, and specialized drivers for individual printers can be had for cost. Might not be suitable for home and hackerish uses in that respect, but business environments would lap it up.
Beg pardon, but Ada Byron/Lovelace's code was written for Babbage's Analytical Engine. Her slacker husband was the poet Lord Byron. Ada married Byron, and did her work with Babbage on the engines. Babbage != Byron; in fact they were some distance apart, though it seems that Byron was fairly supportive of Ada's work, even though at the time the involvement of women in science and mathematics was fairly limited and to some degree discouraged. Ada was herself encouraged by her (cretin) mother to study math, because she (Ada) was getting interested in boys, and her mother (who hated men in general) feared for her virtue. Turned out that Ada liked numbers as much as boys, so her mother did succeed in turning her off men, or at least men who didn't crunch numbers, until her (arranged) marriage to Byron. Ada had three kids and raised them all, whilst working with Babbage.
Babbage himself, not having the advantage of any formal procedures of computer design, never really finished either Engine, since he got distracted by the Difference Engine by the prospects of the Analytical Engine, and the latter was largely doomed by the inadequacy both of mechanical calculators and of machining techniques in England and the world at large of the period (roughly 1830-death, since Babbage never kept up even after being discredited).
Not sure if Ada qualifies as a geek or not; the attributes of geekiness hadn't really been established, and she was required to also be Lady Ada Byron of Lovelace, which is probably like wearing a suit and such. Maybe she was just really smart and good with numbers -- there's not much record of what she herself was like, other than wrt math.
Her programs have all been published, BTW; the one for calculating the seventh Bernoulli number is used in math texts, including the bug and the design flaw. :)
I would infer that an MS strategy to try to undercut the servers would only work if MS marketed to those parts of an academic structure which don't know anything about computers -- which is to say, the administrators, deans, etc., who make the decisions and spend the money and don't actually teach or do much of anything.
Also at issue, of course, is money -- budgets for CSU have been getting cheerfully cut by the CA legislature for some time now, and CS has had its time-to-graduation increase by 25% or so because of restricted class availability (this works out to be more profitable for the institution, which gets paid by enrollment by the gov't plus tuition per student per semester). With some flag-waving, Linux can win out in that respect -- "discounted" can't compete with "free," though the problem is largely the same as competing for prestige with corporate purchasing execs who know little beyond what's written on advertisements.
It's a wholly valid line of discussion. Commercial software has gender-targeting; Linux users get most of their software through their distribution. Thus it's a useful thing to reason through. I don't think it's an especially good idea, but it is a useful thing to consider, esp. wrt the software of interest to women.
FWIW, open source software is much friendlier overall to women -- most of our software makes things and does things -- and those aren't specific to males more than females any more than hammers or screwdrivers. Most of the gender differentiation is based on games, and of our games, we've got a few gore/violence/etc games, and quite a lot of stuff with redeeming qualities like mental exertion and stuff. Not doing too bad there.
For the OSS community as a whole, we've got a long way to go before this stuff can approach gender parity -- and attitudes like yours are, sadly, part of that long way.
FWIW, the MWave modem&sound card was put in a lot of laptops -- IBM doesn't like reproducing engineering. It's pretty much the same as most cards of its type -- basically a DSP, some support circuitry and a driver. It's not even that bad an idea, since a DSP-only modem uses less power and takes up less physical space, both of which are at a premium in laptops. For a long time IBM refused to release the specs to the thing, leading to hacks to make it work in SB compatibility mode. More recently (and the data on this point is confused a bit), IBM seems to have finally recanted and provided some basic specs for talking to the thing. Dale Wick has been working on making the thing work, or at least controlling it a bit, under Linux.
This doubleplusmultisucks.
I saw an external Firewire drive for the first time yesterday (in a Mac computer shop, which I don't frequent, so that's likely why). Looked like an external drive with a big price tag. I'd have preferred to see firewire survive, since it had a lot of potential for network-fabric uses. It concerns me particularly that if Apple stakes much on Firewire, they could be hurt when the Wintel machine lumbers to USB without looking back. While I don't like Apple's patent/copyright tendencies, I like them much better than the Wintel congingent's, and I like having Apple around and doing well. A Wintel-dominated industry is really, really depressing.
Those who make laws in the US are very often former attorneys, or in some cases law enforcement officials or ex-military officers. Granted lawyers are intelligent people, but those who hold legislative office are generally subject to no deep understandings of anything other than bureaucratic and other governmental process. So to put it mildly, they don't seem content with a canonical cover of a set of laws.
But, looked at another way, most legislators have placed their entire faith in their own laws, and have never learned to deal with defiance. Like Aman Hannesy (sp) said to a judge whilst being prosecuted, "Aw judge, your damn laws... the good people don't need 'em and the bad people don't obey em, so what good are they anyway." Theorize: what would happen if everyone, simultaneously, ceased obeying crypto export laws?
Speculation is perfectly fine. It's just hazardous to act on those speculations. Significant distinction.
And I had commercially sensitive data in my email (which would be stupid on a non-POP3 server)
I hope you're not inferring that it's a good idea to pass data through a POP3 server. Not sure if you've encountered this one yet, but POP3 (and most of its kindred) send passwords and mail in the clear, the same way hotmail does. Indeed hotmail would be slightly more secure, since the passwords are likely sent in a POST form, which is mime64-encoded and thus very slightly protected against casual over-shoulder interception. Further, POP is a much more common target for interception since its use is so widespread and the format is quite standardized.
"Secure mail," inasmuch as that can be taken as anything but a contradiction in terms, involves stuff like a secure transmission client, encrypted channels all the way from sender to recipient, storage in encrypted form or on a cryptographic filesystem on a trusted, isolated server, and a secure reception client. At present hardly any such systems exist. The ones that do -- well, they don't run POP3.
As soon as Linus / Alan figgers out how to upgrade an entire kernel from source without rebooting, I'll be happy! ; )
Solaris can do it... they "recommend" going singleuser first, but it amounts basically to a hot-swap kernel. I have no clue how it's done, but I'm jealous too.
I suppose in a HA cluster, you would generally count cluster uptime rather than machine uptime -- maybe then individual kernel upgrades could be discounted so long as the cluster output was still flowing. Hmm.
Even without the merger/buyout consequences, few consumers are prepared to cope with the maneuvering performed by credit companies. The linuxfund APR is a mild (there are much worse) example -- 3.9% for four months, long enough for consumers to get into more debt than they can easily pay off. Then the "introductory" period ends, the rates more than triple, and suddenly many people are left with debts that grow faster than they can pay them back off. Thence they're in thrall to one of the worst kind of creditors one can have.
And, unfortunately, there aren't any fully effective alternatives in the American consumer money tool thing.
That was my interpretation of the posting also. We've seen examples of what happens when MS controls a programming language -- it becomes virtually unusable for any real work. Hence VB and its derivatives. Not that MS even being fully acquitted of infringement in the Java case would lead directly to MS controlling Java; more likely they'd just be free to pollute Java and try to emasculate it in favor of stuff that doesn't work cross-platform. Which would be unfortunate, because Java is a fairly neat bunch of things which haven't been done together in a while.
Ah, here 'twas: http://www.inf.bme.hu/~dancsi/USE!/
I bought a copy of xingenc when it was released for Linux (impulse purchase). The speed was quite good, and the speed/quality ratio was also quite ideal. A while back I did a by-ear comparison of bladeenc, xing and 8hz' encode -- bladeenc and encode were both good quality, their perceptible flaws were some slight muddling when both high and low freqs were occuring simultaneously in the music (which is a waveform encoding issue we're not going to get away from). Xing's quality was perceptibly lower, audibly distorting both extremes of the frequency range (plus chopping the ones higher than you're supposed to be able to hear). That was all at 128kbps -- at 160kbps and above, I couldn't make out any difference.
OTOH, the Xing encoder supports VBR encoding, which did seem to increase the perceived quality quite a bit. The feature is available under Linux; the files are a bit bigger with a mid-to-high quality settings, but the result sounds nice.
Somewhere there was an actual frequency-analysis report of a bunch of mp3 encoders, including all three of the above. Now I've forgotten the URL.
2+2=5 was one of the precepts used in the interrogation (reeducation, brainwashing, whatever you'd rather call it) of Winston in 1984. Not the main criteria, but one of them.
"Don't do it to me, do it to Julia!"There are a few high school English classes that cover 1984. There are a million or so more that don't. Pity, because it's a more useful text for your average public school student than much of what gets covered there.
In any case, a server should be able to survive on zero writeable disk space. It's acceptable, if undesirable, for operations that involve writing to the disk (which does not include static HTML, that being all I saw on the test site), to fail semi-gracefully.
:)
Under most decent OSes, writes to a disk-full file on a normal filesystem will get bufferred in RAM until free space opens up; when the RAM's full, it has to start discarding those buffers, once it's pared down the cache and forced some processes out to swap. Maybe that's involved, maybe not. The suggestion that the machine has a lot of logging turned on so they could benefit from successful crashes has merit, though -- although I'd be logging it to a different machine with a huge pile of disks that was also logging all the net traffic. Oh well. "Poor MS."
Pray don't confuse firewalls with security. Machines behind a firewall are only as secure (from the big bad net, assuming they're connected to it, as most firewalls are) as the TCP/IP stacks and services that answer on ports accessible through the firewall and/or its sockets. Most firewalls that I've had experience with have closed off all ports except the ones that people needed to use from the outside -- and half the time that included stuff like pop[23], imap, smtp, etc., and on which the servers answering those ports tended to be way behind on their updates because people had this sense of security lent by the firewall.
Also, firewalls don't work from people who can emit packets from inside your firewall -- and that's surprisingly easy to do, either through coercion of the firewall box's network stack, compromise of a machine behind the firewall through some open port, or simply being behind the firewall in the first place (as in many corporate environments). If a firewall is configured to permit connections to ports 22 (ssh) and 443 (SSL http), there's no particular reason why an attacker can't arrange for a root shell to answer on one of those ports, and with most network installations no one would be the wiser.
#2 has an implementation benefit as well -- as the Echelon paper highlighted, none of the major US organizations responsible for the widespread electronic interferenec with our privacy have the CPU power necessary for widespread brute-force attacks on our ciphers. Therefore, one good way to oppose the US government's invasions of our privacy is to encrypt as great a volume of our traffic as possible, thus presenting the FBI, NSA etc. with so much enciphered traffic that they can't hope to ascertain what in it was wortht he CPU time. Crypto makes for good privacy against wiretapping in its own rights, and one can help thwart those who would trespass on our privacy simply by participating in overwhelming their CPUs.
:)
In other words, encrypt everything, whether it needed it or not. By a happy coincidence, the internet is wonderful for that.
After perusing through them, I find also that the message boards attached to the poll reflect a level of intelligence, respect and tact that makes slashdot look like a graduate debate seminar. Between the Theologically Impaired and the Single-Issue Media Potatoes, it's a large and gratuitous mess.
Without endorsing any specific approach, when you do your architecture design, look for opportunities to make it distributable. In general it's always cheaper to buy ten average PCs than it is to buy a single quad-xeon or Sun E10k or such -- though the job may not require that much if you do your engineering right. In the scenario you've described, look for use of multiple POP/IMAP servers sharing a common data repository (or, if it's not essential to have everyone share the same disk, divide that duty up also). Check out the high-availability Linux project for useful tools. Even if you wind up building things on one big server rather than several littler ones, leave room in your software setup so that you can distribute the load later on if need be without major reengineering.
And stay the hell away from Exchange. The degree of pain to be had from that pile of dung defies description, and the pain seems to grow exponentially rather than linearly as demand rises.