Anyone worth half a karma point here will recognise 16384 as a power of two.
In my years of software development, numbers like this jump out at you, usually while debugging something that has crashed due to overwriting something, and suspicious powers of two just scream 'BUG' at me.
Perhaps this move to manufacturing has simply been caused by microsoft not allocating enough bits in the build number, and one more recompile has tripped the manufacturing release...
struct BuildNumber {
int IncrementalVersion : 14;
int ReleaseToManufacturing : 1;
int FinallyBugFree : 1; }
(and if this really is the source code, we'll have to wait until release 32768 for a bug free version, assuming we don't hit -32768 first)
Greebo had spent an irritating two minutes in that box. Technically, a cat locked in a box may be alive or it may be dead. You never know until you look. In fact, the mere act of opening the box will determine the state of the cat, although in this case there were three determinate states the cat could be in: these being Alive, Dead, and Bloody Furious.
Shawn dived sideways as Greebo went off like a Claymore mine.
"Don't worry about him," said Magrat dreamily, as the elf flailed at the maddened cat. "He's just a big softy."
Over here in the UK, the high definition video providers are locking their signals up tightly inside HDCP-encrypted HDMI links, and not allowing you access to the high definition signals in any way. No component outputs, no RGB, no VGA, nada. Very annoying.
In India, it is fairly common practice for relatives to get family members declared officially dead so that they can get their hands on their property.
This unfortunate state of affairs is the reason that Lal Bihari from Uttar Pradesh, India, formed the Association for Dead People, to draw attention to the problem. Lal, who discovered that he was officially dead in 1976, took eighteen years to prove to the indian government that he was still very much alive.
He has since been awarded the 2003 Ig Nobel Peace Prize for his exploits, for "waging a lively posthumous campaign against bureaucratic inertia and greedy relatives".
The UK Telephone preference service states in their FAQ that Residential or Sole Trader numbers do not need to be renewed, however corporate numbers (for corporations who do not want to be bothered by other corporations making unsolicited sales calls) do need to renew their CTPS subscription every year.
Interestingly, they ask that if you move, but maintain the same telephone number, to contact them. It may be implied from this (although I have found no evidence) that ceased telephone numbers are removed from the list in some automated telco-DMA manner, which would make sense (but since when as making sense had anything to do with real life?)
Interestingly, the Telephone Preference Service is a supression list service maintained by the direct marketing association, but is a regulatory requirement specified in the Privacy and Electronic Communcations (EC Directive) Regulations 2003 which states that it is OFCOM's (the regulator) responsibility to maintain such a list.
Confused? I am... (although having been a TPS subscriber on all my lines for some years, it does appear to work).
Unfortunately for the O.P. the DMCA will rule his watching of unencrypted video feeds a violation, because he is circumventing a protection mechanism.
The DMCA and it's supporters have shown time and time again that it does not matter how lame and technically incompetent a protection measure is, it merely has to be 'broken' to incur the wrath of the law. Publishing the technical details is probably instant guilt in terms of 'trafficking in circumvention technology' or whatever the legalese is.
The early analogue cable TV transmission systems in the UK used to transmit premium channels unencrypted (and then moved to using some fairly trivial to counter sync-mangling) which simply required a tuner that could see outside of the usual UHF 21-68 band to view. While the frequencies used on the actual cable networks were sufficiently out-of-range of normal 'terrestrial' channels, the company actually supplied an add-on box clamped to the back of the cable receiver which would downshift all of the cable channels so that the normal 'terrestrial' channels carried on the cable service were tunable by 'normal' TV's. If the TV had a tuner which could see outside the usual UK 21-68 band (Channel 21 is 471.25MHz, 68 is 847.25MHz) then the channels are there for your viewing pleasure regardless of what the cable receiver thinks you are entitled to see.
All the article does is move this into the digital realm with QAM.
Minus several million points out of ten to comcast for not encrypting traffic here.
My initial suspicion was that this company is trying to commit suicide.
However, after engaging the brain for a microsecond, I suspect what they are trying to do is get themselves bought out, because that result is probably cheaper in the long run to one of the big DRM users out there (mm. surprised they didn't sue Sony/Disney)
Otherwise I read the case like this: I don't pay you to get your car keys from you in order to steal your car. I don't steal your car. I don't even know where your car is, and have no intention of stealing it, but I'm guilty of not using the official theft-prevention technology (i.e. your keys) to not steal it. I think that makes about as much sense as this lawsuit.
Polycarbonate discs are ridiculously cheap, and quite durable
A polycarbonate disc with any form of chip inserted into the structure will weaken the disc, leading to cracks, and exploding discs. DVDs are spun at various speeds, faster in some drives than others. Perhaps if the chip could be programmed to detect unauthorised players and detonate in them... no, I don't want to give the DVD CCA any more stupid ideas...
A single chip will not do either, because it will create an imbalance, which the mechanism of the drive (which is of course, as cheap and feeble as the manufacturer can realistically make it) would have to cope with, so there would have to be two chips, placed exactly 180-degrees around the disc. That would also contribute towards straight-line fractures. The only place a single chip could be placed to avoid that problem is right in the centre of the disc, where there is, unfortunately, already a hole.
Retail stores in the UK cut down on DVD theft by keeping DVDs behind the counter, leaving only the cases on display. The cases also contain assorted anti-theft devices which react with the radio-based scanners at the shop entrance.
I rarely buy CD's/DVD's at shops anyway; they are too expensive. I tend to buy via the net. (One boxed set of CD's recently cost me £38 including tax, import duty, and carriage from the USA, compared to £79 in the high street. These particular discs were marked 'Made by EMI Swindon' and had been reduced by over 50% by shipping them across the atlantic twice. Go figure.
But let's say you have to open the device, and the case is designed to break when that happens.
Thanks, just another excuse for the manufacturers to seal their devices in goo to stop us taking them apart and finding out how they work (so we can install Linux on them, etc.).
The UK government has recently instructed magistrates and judges not to jail non-violent offenders where possible, due to lack of space in the countries' already crowded prisons.
While the threat of jail is still there, the chances of anyone actually getting a custodial sentence for such crimes is virtually non-existant, when even major crime gets punished with fines and community service.
So, yet another UK law that looks good on paper, but will be as effective as the USA CAN-SPAM laws.
Come on now, it's a new release of BloatCorp's software. Simply take the minimum system requirements of the previous version, add 10% for each use of the word 'More' or 'Improved' in the product blurb, then multiply by the ratio of the retail price of the new and old versions.
The UK manufactures banknotes for a number of countries other than itself (we make Euro coins too, despite not actually being in the 'EuroZone'). The USA could outsource it's money production to a cheaper country, or a country already set up to print secure currency.
If the notes are redesigned, they would hopefully be designed to the latest anti-forgery standards, with holograms, embedded or overprinted strips (reflective, temperature sensitive, holographic, interwoven etc.), partial designs on opposite sides (so that if both sides are printed even slightly off-centre, the note looks completely wrong when held up to light)
Besides, machinery like that can pay for itself, literally;-)
> it just exposes the stupidity of the programmer.
Indeed; The programmer may have printf("%d\n",(int)log10(0.0001)); and been suitably suprised.
The oddity here is that with the hardware using 80-bit, and the software 64-bit, the particular value generated is inconsistent if the apparently same operation is carried out on it twice. FIST generates -4; FST generates -4.000, FISTP generates -3 or -4 depending on the round/chop bit in the FPCW. The (microsoft) inconsistency is that the int = (int)double cast does not behave consistently, which will further confuse the hapless programmer.
My programming background is in bits, registers, and I/O, not doubles, hence I had to debug this particular instance for the programmer concerned. I'd have done it in ASM, and kept the whole thing in 80-bit, but that is why I don't do the mathematical stuff in the software, just the bit bashing..;-)
Also due to the lack of proofreading/previewing, the 200000000010 should have had a ^ inserted, as in 1.0000000000... x 2^00000000010 Pity slash's allowed HTML does not include <SUP> tags..
Bugger! Missed 'preview' and pressed 'submit'. A lot of the tabs got messed up.
I was going to add: d = log10(0.0001); i = (int)d; printf("%d\n",i); i = (int)d; printf("%d\n",i); ...will generate different results at each printf call, confusing the novice programmer to the point of despair;-)
I wrote this article for our in-house programmers some time ago to illustrate just this problem...
Quantum mechanics on the floating-point stack
This article describes a potential pitfall for visual C 6.0 programmers when using floating point arithmetic. This does not occur with GCC, even on the same platform, because the (int) typecast is sufficiently different.
Why 'Quantum'? The third principle of quantum mechanics is "The act of observing something happening causes a change in the thing being observed." This holds true in the examples given in this article.
The problem stems from the way floating point numbers are handled by the hardware in the Intel-compatible floating point units. Externally to the floating point unit, floating point values are stored in 32-bit (single) or 64-bit (double) values. However, internally, they are stored in an 80-bit format, as below:
Both the single and double precision floats also have an implied leading-1 on the mantissa which is not stored in the actual data, whereas the internal float does not.
The value -4.0 is stored in a double as (hex) 00 00 00 00 00 00 10 C0, when this is written in binary, in the byte order of the diagram above, this works out as 1 10000000001 00000000000000000000000000000000000000000000000000 000000 The exponent is biased, which means that it has a value added to it to provide for negative exponents. In double precision, that bias value is 1023. Expressed in binary, that is 01111111111, so the exponent's real value is 10000000001 - 01111111111 = 00000000010. The mantissa has an implied leading 1, and the decimal point appears between that leading 1 and the rest of the mantissa, so the value above is negative (the sign bit is set) 1.0000000000... x 200000000010 = -1.0 x 2^2 = -4.0
So far, so good. However the problem lies in functions such as log10(), pow() and so forth that generate floating point results which suffer from accuracy problems. On the intel architecture, these accuracy problems only manifest themselves at the extreme end of the 80-bit floating point format, so for example log10(0.0001) returns -3.99999999999999997e+0000 in the 80-bit format. When this value is stored in a double, via a FST(P) (floating store (and pop)) instruction, the value is rounded to 64-bit accuracy, and results in the bit patten above.
If you copy this value to an integer, you will get the correct result, which is -4.
The error creeps in when you copy the value to an integer directly from the floating point stack, because the value on the floating point stack is still -3.99999999999999997e+0000, and the '=(int)' typecast always chops the decimal places off. The processor converts floating point to integer via a FIST(P) (floating integer store (and pop)) instruction, and the function which the compiler uses to do the cast sets the floating point control word to chop, rather than round, the result. This function, called __ftol is automatically called by the Visual C compiler to do the conversion. It is slow and cumbersome, but it does the job to the specification required. However, you can get different results, depending on when the function is called, because the compiler generates different code depending on how you use floating point results. In all of the following examples, the stack management code has been removed for clarity.
In the instance: d = log10(0.0001);
i = (int)d; the microsoft compiler generates the following code: push 3F1A36E2h ; Push high 32-bits of 0.0001 (double) onto stack
push 0EB1C432Dh ; Push low 32-
ISO9660 has version numbers, the syntax is the same as per VMS. The actual space reserved in ISO 9660 level 1 is 31 characters for the name, and six further characters for ';NNNNN' where NNNNN is the version number. If you use mkisofs to create an ISO and examine the contents, you'll find the version number in there, with all of the base ISO names in upper case.
Various extensions to ISO9660 (including ISO9690:1999) may cause the version numbers to not be omitted, but doing so violates ISO9660 (which is usually not a huge problem, except for really dumb devices).
The main problem with version numbers in VMS was that discs were a lot smaller when VMS was more common. Consequently, runaway applications would generate enormous numbers of versions, and system managers would typically issue a SET DIRECTORY/VERSION_LIMIT=5 DISK$FOO:[000000...]*.DIR (not entirely sure if I got the syntax right, I have a MicroVAX under my desk, but it has not been switched on for six years). Also, when discs got full, the desperate sysadmin would probably PURGE/KEEP=3 across the entire disk to trim excess junk.
As the System V.2 administrator's guide says: Making files is easy under the UNIX operating system. Therefore, users tend to create numerous files using large amounts of file space. It has been said that the only standard thing about all UNIX systems is the message-of-the-day telling users to clean up their files.
Linux 2.0.35 had a patch allowing 2048-byte sectors on SCSI devices; handy if you had a Fujitsu Magneto-optical drive with a capacity of 640Mb/disc or more, which used 2k sectors.
As the patch was done 'properly', a couple of tweaks of some constants and a recompile (if it isn't a run-time parameter already) should enable 4k sectors, 8k sectors, even 1Mb sectors, if you really want to go there.
Here in the UK, we have been shipping linux systems both within the country and around the world.
Our customers in some cases demand Dell hardware. In some other cases they require three years of next-business-day hardware support, and about the only manufacturer from which this is available at a reasonable price (in places like downtown Bangalore, for instance) is Dell.
Dell have been extremely unhelpful when it comes to supplying systems with linux preinstalled. For an example, see Exhibit A showing various Dell Precision systems, clearly showing a choice of operating systems. Then click on the Precision 380 (Exhibit B) which conveniently removes the choice.
Should you actually want to customise and buy a system, you get a screen looking like Exhibit C clearly providing you a choice between 'Genuine Windows(R) XP Professional, SP2(NTFS)(+Media) [Included in Price]' and exactly the same thing installed on FAT32.
This is, perhaps, Dell UK's definition of 'Choice'.
Now, why did I find the test challenging but enjoyable, but the others found it impossible?
Probably for the same reason that some people like Su Doku, and others refuse to even consider looking at it. (I'm in the former category, especially the so called 'Deadly' version with the shapes that add up to a specific value with no starting numbers)
Some people enjoy a challenge, others do not. People these days are being bred not to enjoy challenges, and to ask 'Why should I?' when presented with one.
I've had enough. Please stop this planet, I want to get off.
Slashdot Mirror
Anyone worth half a karma point here will recognise 16384 as a power of two.
In my years of software development, numbers like this jump out at you, usually while debugging something that has crashed due to overwriting something, and suspicious powers of two just scream 'BUG' at me.
Perhaps this move to manufacturing has simply been caused by microsoft not allocating enough bits in the build number, and one more recompile has tripped the manufacturing release...
struct BuildNumber
{
int IncrementalVersion : 14;
int ReleaseToManufacturing : 1;
int FinallyBugFree : 1;
}
(and if this really is the source code, we'll have to wait until release 32768 for a bug free version, assuming we don't hit -32768 first)
From Wikipedia:
I guess the Internation Obfuscated Ada Code Competition is a non-starter then.
(Notes that ioacc.org is unregistered)
Compare IOCCC
Over here in the UK, the high definition video providers are locking their signals up tightly inside HDCP-encrypted HDMI links, and not allowing you access to the high definition signals in any way. No component outputs, no RGB, no VGA, nada. Very annoying.
> and 20/5 internet
Personally, I'd prefer 24/7
You play cricket?
In India, it is fairly common practice for relatives to get family members declared officially dead so that they can get their hands on their property.
This unfortunate state of affairs is the reason that Lal Bihari from Uttar Pradesh, India, formed the Association for Dead People, to draw attention to the problem. Lal, who discovered that he was officially dead in 1976, took eighteen years to prove to the indian government that he was still very much alive.
He has since been awarded the 2003 Ig Nobel Peace Prize for his exploits, for "waging a lively posthumous campaign against bureaucratic inertia and greedy relatives".
News coverage in New York Times,
The UK Telephone preference service states in their FAQ that Residential or Sole Trader numbers do not need to be renewed, however corporate numbers (for corporations who do not want to be bothered by other corporations making unsolicited sales calls) do need to renew their CTPS subscription every year.
Interestingly, they ask that if you move, but maintain the same telephone number, to contact them. It may be implied from this (although I have found no evidence) that ceased telephone numbers are removed from the list in some automated telco-DMA manner, which would make sense (but since when as making sense had anything to do with real life?)
Interestingly, the Telephone Preference Service is a supression list service maintained by the direct marketing association, but is a regulatory requirement specified in the Privacy and Electronic Communcations (EC Directive) Regulations 2003 which states that it is OFCOM's (the regulator) responsibility to maintain such a list.
Confused? I am... (although having been a TPS subscriber on all my lines for some years, it does appear to work).
Unfortunately for the O.P. the DMCA will rule his watching of unencrypted video feeds a violation, because he is circumventing a protection mechanism.
The DMCA and it's supporters have shown time and time again that it does not matter how lame and technically incompetent a protection measure is, it merely has to be 'broken' to incur the wrath of the law. Publishing the technical details is probably instant guilt in terms of 'trafficking in circumvention technology' or whatever the legalese is.
The early analogue cable TV transmission systems in the UK used to transmit premium channels unencrypted (and then moved to using some fairly trivial to counter sync-mangling) which simply required a tuner that could see outside of the usual UHF 21-68 band to view. While the frequencies used on the actual cable networks were sufficiently out-of-range of normal 'terrestrial' channels, the company actually supplied an add-on box clamped to the back of the cable receiver which would downshift all of the cable channels so that the normal 'terrestrial' channels carried on the cable service were tunable by 'normal' TV's. If the TV had a tuner which could see outside the usual UK 21-68 band (Channel 21 is 471.25MHz, 68 is 847.25MHz) then the channels are there for your viewing pleasure regardless of what the cable receiver thinks you are entitled to see.
All the article does is move this into the digital realm with QAM.
Minus several million points out of ten to comcast for not encrypting traffic here.
otherwise, it's old news.
My initial suspicion was that this company is trying to commit suicide.
However, after engaging the brain for a microsecond, I suspect what they are trying to do is get themselves bought out, because that result is probably cheaper in the long run to one of the big DRM users out there (mm. surprised they didn't sue Sony/Disney)
Otherwise I read the case like this: I don't pay you to get your car keys from you in order to steal your car. I don't steal your car. I don't even know where your car is, and have no intention of stealing it, but I'm guilty of not using the official theft-prevention technology (i.e. your keys) to not steal it. I think that makes about as much sense as this lawsuit.
Polycarbonate discs are ridiculously cheap, and quite durable
A polycarbonate disc with any form of chip inserted into the structure will weaken the disc, leading to cracks, and exploding discs. DVDs are spun at various speeds, faster in some drives than others. Perhaps if the chip could be programmed to detect unauthorised players and detonate in them... no, I don't want to give the DVD CCA any more stupid ideas...
A single chip will not do either, because it will create an imbalance, which the mechanism of the drive (which is of course, as cheap and feeble as the manufacturer can realistically make it) would have to cope with, so there would have to be two chips, placed exactly 180-degrees around the disc. That would also contribute towards straight-line fractures. The only place a single chip could be placed to avoid that problem is right in the centre of the disc, where there is, unfortunately, already a hole.
Retail stores in the UK cut down on DVD theft by keeping DVDs behind the counter, leaving only the cases on display. The cases also contain assorted anti-theft devices which react with the radio-based scanners at the shop entrance.
I rarely buy CD's/DVD's at shops anyway; they are too expensive. I tend to buy via the net. (One boxed set of CD's recently cost me £38 including tax, import duty, and carriage from the USA, compared to £79 in the high street. These particular discs were marked 'Made by EMI Swindon' and had been reduced by over 50% by shipping them across the atlantic twice. Go figure.
But let's say you have to open the device, and the case is designed to break when that happens.
Thanks, just another excuse for the manufacturers to seal their devices in goo to stop us taking them apart and finding out how they work (so we can install Linux on them, etc.).
Don't give them ideas!
The UK government has recently instructed magistrates and judges not to jail non-violent offenders where possible, due to lack of space in the countries' already crowded prisons.
While the threat of jail is still there, the chances of anyone actually getting a custodial sentence for such crimes is virtually non-existant, when even major crime gets punished with fines and community service.
So, yet another UK law that looks good on paper, but will be as effective as the USA CAN-SPAM laws.
Come on now, it's a new release of BloatCorp's software. Simply take the minimum system requirements of the previous version, add 10% for each use of the word 'More' or 'Improved' in the product blurb, then multiply by the ratio of the retail price of the new and old versions.
The UK manufactures banknotes for a number of countries other than itself (we make Euro coins too, despite not actually being in the 'EuroZone'). The USA could outsource it's money production to a cheaper country, or a country already set up to print secure currency.
;-)
If the notes are redesigned, they would hopefully be designed to the latest anti-forgery standards, with holograms, embedded or overprinted strips (reflective, temperature sensitive, holographic, interwoven etc.), partial designs on opposite sides (so that if both sides are printed even slightly off-centre, the note looks completely wrong when held up to light)
Besides, machinery like that can pay for itself, literally
> it just exposes the stupidity of the programmer.
;-)
Indeed; The programmer may have printf("%d\n",(int)log10(0.0001)); and been suitably suprised.
The oddity here is that with the hardware using 80-bit, and the software 64-bit, the particular value generated is inconsistent if the apparently same operation is carried out on it twice. FIST generates -4; FST generates -4.000, FISTP generates -3 or -4 depending on the round/chop bit in the FPCW. The (microsoft) inconsistency is that the int = (int)double cast does not behave consistently, which will further confuse the hapless programmer.
My programming background is in bits, registers, and I/O, not doubles, hence I had to debug this particular instance for the programmer concerned. I'd have done it in ASM, and kept the whole thing in 80-bit, but that is why I don't do the mathematical stuff in the software, just the bit bashing..
Also due to the lack of proofreading/previewing, the 200000000010 should have had a ^ inserted, as in 1.0000000000... x 2^00000000010
Pity slash's allowed HTML does not include <SUP> tags..
Bugger! Missed 'preview' and pressed 'submit'.
...will generate different results at each printf call, confusing the novice programmer to the point of despair ;-)
A lot of the tabs got messed up.
I was going to add:
d = log10(0.0001);
i = (int)d;
printf("%d\n",i);
i = (int)d;
printf("%d\n",i);
Quantum mechanics on the floating-point stack
This article describes a potential pitfall for visual C 6.0 programmers when using floating point arithmetic. This does not occur with GCC, even on the same platform, because the (int) typecast is sufficiently different.
Why 'Quantum'? The third principle of quantum mechanics is "The act of observing something happening causes a change in the thing being observed." This holds true in the examples given in this article.
The problem stems from the way floating point numbers are handled by the hardware in the Intel-compatible floating point units.
Externally to the floating point unit, floating point values are stored in 32-bit (single) or 64-bit (double) values. However, internally, they are stored in an 80-bit format, as below:
single float: Sign <--8-bit-exponent--> <--23-bit-mantissa-->
double float: Sign <-11-bit-exponent--> <--52-bit-mantissa-->
x87 register: Sign <-15-bit-exponent--> <--64-bit-mantissa-->
Both the single and double precision floats also have an implied leading-1 on the mantissa which is not stored in the actual data, whereas the internal float does not.
The value -4.0 is stored in a double as (hex) 00 00 00 00 00 00 10 C0, when this is written in binary, in the byte order of the diagram above, this works out as
1 10000000001 00000000000000000000000000000000000000000000000000 000000
The exponent is biased, which means that it has a value added to it to provide for negative exponents. In double precision, that bias value is 1023. Expressed in binary, that is 01111111111, so the exponent's real value is 10000000001 - 01111111111 = 00000000010.
The mantissa has an implied leading 1, and the decimal point appears between that leading 1 and the rest of the mantissa, so the value above is negative (the sign bit is set) 1.0000000000... x 200000000010 = -1.0 x 2^2 = -4.0
So far, so good. However the problem lies in functions such as log10(), pow() and so forth that generate floating point results which suffer from accuracy problems. On the intel architecture, these accuracy problems only manifest themselves at the extreme end of the 80-bit floating point format, so for example log10(0.0001) returns -3.99999999999999997e+0000 in the 80-bit format. When this value is stored in a double, via a FST(P) (floating store (and pop)) instruction, the value is rounded to 64-bit accuracy, and results in the bit patten above.
If you copy this value to an integer, you will get the correct result, which is -4.
The error creeps in when you copy the value to an integer directly from the floating point stack, because the value on the floating point stack is still -3.99999999999999997e+0000, and the '=(int)' typecast always chops the decimal places off. The processor converts floating point to integer via a FIST(P) (floating integer store (and pop)) instruction, and the function which the compiler uses to do the cast sets the floating point control word to chop, rather than round, the result. This function, called __ftol is automatically called by the Visual C compiler to do the conversion. It is slow and cumbersome, but it does the job to the specification required. However, you can get different results, depending on when the function is called, because the compiler generates different code depending on how you use floating point results. In all of the following examples, the stack management code has been removed for clarity.
In the instance:
d = log10(0.0001);
i = (int)d;
the microsoft compiler generates the following code:
push 3F1A36E2h ; Push high 32-bits of 0.0001 (double) onto stack
push 0EB1C432Dh ; Push low 32-
ISO9660 has version numbers, the syntax is the same as per VMS. The actual space reserved in ISO 9660 level 1 is 31 characters for the name, and six further characters for ';NNNNN' where NNNNN is the version number. If you use mkisofs to create an ISO and examine the contents, you'll find the version number in there, with all of the base ISO names in upper case.
Various extensions to ISO9660 (including ISO9690:1999) may cause the version numbers to not be omitted, but doing so violates ISO9660 (which is usually not a huge problem, except for really dumb devices).
The main problem with version numbers in VMS was that discs were a lot smaller when VMS was more common. Consequently, runaway applications would generate enormous numbers of versions, and system managers would typically issue a SET DIRECTORY/VERSION_LIMIT=5 DISK$FOO:[000000...]*.DIR (not entirely sure if I got the syntax right, I have a MicroVAX under my desk, but it has not been switched on for six years). Also, when discs got full, the desperate sysadmin would probably PURGE/KEEP=3 across the entire disk to trim excess junk.
As the System V.2 administrator's guide says:
Making files is easy under the UNIX operating system. Therefore, users tend to create numerous files using large amounts of file space. It has been said that the only standard thing about all UNIX systems is the message-of-the-day telling users to clean up their files.
[caveman@marvin caveman]$ ls -l /etc/motd /etc/motd
crw-r--r-- 1 root root 1, 9 Apr 21 2006
That should stop the buggers logging in.
It's probably related to that dastardly 'kibi', 'mibi' and 'gibi'- prefix plan.
I mean, we all know what a kilobyte is. And by extension, know what a megabyte and a gigabyte is.
We also know that marketeers deliberately do not know what they are, and should be shot on sight. (Where's Cheney when you need him?)
Linux 2.0.35 had a patch allowing 2048-byte sectors on SCSI devices; handy if you had a Fujitsu Magneto-optical drive with a capacity of 640Mb/disc or more, which used 2k sectors.
As the patch was done 'properly', a couple of tweaks of some constants and a recompile (if it isn't a run-time parameter already) should enable 4k sectors, 8k sectors, even 1Mb sectors, if you really want to go there.
Our customers in some cases demand Dell hardware. In some other cases they require three years of next-business-day hardware support, and about the only manufacturer from which this is available at a reasonable price (in places like downtown Bangalore, for instance) is Dell.
Dell have been extremely unhelpful when it comes to supplying systems with linux preinstalled. For an example, see Exhibit A showing various Dell Precision systems, clearly showing a choice of operating systems. Then click on the Precision 380 (Exhibit B) which conveniently removes the choice.
Should you actually want to customise and buy a system, you get a screen looking like Exhibit C clearly providing you a choice between 'Genuine Windows(R) XP Professional, SP2(NTFS)(+Media) [Included in Price]' and exactly the same thing installed on FAT32.
This is, perhaps, Dell UK's definition of 'Choice'.
Probably for the same reason that some people like Su Doku, and others refuse to even consider looking at it. (I'm in the former category, especially the so called 'Deadly' version with the shapes that add up to a specific value with no starting numbers)
Some people enjoy a challenge, others do not. People these days are being bred not to enjoy challenges, and to ask 'Why should I?' when presented with one.
I've had enough. Please stop this planet, I want to get off.