Year 2038 Problem (Devs?)
Year 2038 Problem (Devs?) |
|
Message boards : Number crunching : Year 2038 Problem (Devs?)
| Author | Message |
|---|---|
|
There's a problem with Unix timestamps in the year 2038, have a read of the Year 2038 problem Wikipedia article | |
ID: 249574 |  | |
|
Program developers will continue using it and start to panic in 2037 | |
| ID: 249587 | | |
Program developers will continue using it and start to panic in 2037 ha! that wouldn't surprise me The OS/420 guys will have scrapped the century digits by then and be happy to have another 62 years left. ah yes, i remember that page, and the rather obvious year 10,000 problem too :p i really fail to see why things aren't done properly in this day and age where memory isn't a problem anymore (it was excusable in the early days) i mean, look at the IP address space, IPv4 has 2^32 = 4,294,967,296 possible addresses (but some are reserved, so about 3,900,000,000/4,000,000,000 in practice) which was LOADS when IPv4 was born IPv6 has 2^128 ! which is such a huge number that i'm not going to post it here, but in comparison, maybe a diagram will help To make that diagram to scale, imagine the IPv4 address space is the 1.6-inch square above. In that case, the IPv6 address space would be represented by a square the size of the solar system. the MAC address space is even bigger, so if the network guys can realise that it's better to have too many than too few, why can't the software people learn? just to clear some confusion, this is a problem for boinc, because boinc uses unix timestamps to record times/dates in various places (take a look in the client_state.xml file) if like me you're not very good at reading unix timestamps, there's a converter available | |
| ID: 249607 | | |
|
The date is stored as a either 16 or 32 bit integer (I'm not sure which). By the time the "problem" will actually cause a problem, most of the computers in use now will be retired to that great computer farm in the sky! Haha! The newer computers that are designed would have two options, either keep the same data size for the date but switch to a new "reference date", or use a larger data size to store it. | |
| ID: 249612 | | |
Program developers will continue using it and start to panic in 2037 Using a 64-bit value introduces a new problem date in about 290 billion years, on Sunday, December 4, 292,277,026,596 15:30:08 UTC. This problem is not, however, widely regarded as a pressing issue. ____________ Account Frozen... | |
| ID: 249618 | | |
The date is stored as a either 16 or 32 bit integer (I'm not sure which). By the time the "problem" will actually cause a problem, most of the computers in use now will be retired to that great computer farm in the sky! Haha! The newer computers that are designed would have two options, either keep the same data size for the date but switch to a new "reference date", or use a larger data size to store it. i'm just pointing out the fact that if this was done at a software level, then there wouldn't need to be any work arounds | |
| ID: 249619 | | |
Especially considering that's on the order of thirty times longer than the Sun is expected to last. ____________   | |
| ID: 249702 | | |
well it'll engulf the earth before that (stars grow when they get old, and become red giants) and people will be living far away by then (unless we're extinct :S) | |
| ID: 249718 | | |
i really fail to see why things aren't done properly in this day and age where memory isn't a problem anymore (it was excusable in the early days) The UNIX time stamp format was developed in "the early days" and is now used in many places - just like the 2 digit date. However I predict a much smoother transition this time. No freaks sitting out in the desert with guns waiting for the rest of society to crumble. As was mentioned above, 64 bit operating systems should be in widespread use by 2038 and this will pretty much be a non-problem. ____________ A member of The Knights Who Say NI! For rankings, history graphs and more, check out: My BOINC stats site | |
| ID: 249723 | | |
The date is stored as a either 16 or 32 bit integer (I'm not sure which). By the time the "problem" will actually cause a problem, most of the computers in use now will be retired to that great computer farm in the sky! Haha! The newer computers that are designed would have two options, either keep the same data size for the date but switch to a new "reference date", or use a larger data size to store it. After thinking about this a little, I can see what you are trying to say, but if you store the date using the same data size, and your program said to get the date and it read a date of 000000... how would your software know which date this refered to since any particular date that consists of all zeros would reappear every 68 years. It could mean 1970, 2038, 3006, etc. How could you tell? This is the same problem that was seen in the last of the 90's for programs that stored date information as two digits instead of the full 4 digit year. If you saw "06", is it 2006, 1906, 1806, 3706, etc. You couldn't know. For your solution to work, there would have to be certain assumptions made. Basically, right now a computer looks at the date and interprets it as seconds from 01/01/70. The computers of the future would have to look at the same number and interpret it as being seconds from 01/01/2038. But that brings up another problem which is at that time all software, bioses, etc. that depended on the stored date would have to be upgraded exactly at midnight on that date or else it would be reporting the wrong date. ____________ Jim Join GQFCrunchers Problems? Check the Boinc Wiki Or NEW Enhanced FAQ | |
| ID: 249811 | | |
... But that brings up another problem which is at that time all software, bioses, etc. that depended on the stored date would have to be upgraded exactly at midnight on that date or else it would be reporting the wrong date. exactly, i was trying to think how to explain that concept clearly, but you hit the nail on the head i just find it ignorant of developers/designers to use the easy way (2 digit years for example) which causes many other problems, rather than using a better solution from the start (it's easier to use a different system from the start rather than try to change an existing one) maybe some kind of extensible date/time format needs to be invented, one that doens't need a fixed length value (it can get longer) so instead of 1 second being 000...0000001 it would just be 1 10 would be 10, 10, 100 etc. and software just parses the number in the reverse order, with the digit at the far right always representing a particular unit (like seconds for example) and it works out the time/date from right to left, and keeps going (reading additional digits) until it reaches the "start tag" as it were (such as in XML) | |
| ID: 249816 | | |
|
To quote "John Titor" (an interesting read in & of itself about him): | |
| ID: 249820 | | |
"UNIX has a problem in 2038..." Not 100% sure of my first fact, but I understand motherboard and BIOS suppliers are looking to move to a new format which does away with the BIOS in the next couple of years. The second point is speculative, but likely to become fact:- I cannot remember when the change between 8 bit and 16 bit computing occurred, but I assume this was around 1984/85/86. The change from 16 bit to 32 bit computing occurred around 1994/95. We are currently moving between 32 bit and 64 bit, likely to become mainstream in 2007/08. This suggests the timing for the change over, as a time line is ... The move from 8/16 bit to 16/32 bit took, shall we say, 10 years? The move from 16/32 bit to 32/64 bit appears to be occurring 12 years after the last change. Speculating ... In about 14 - 15 years we will be addressing a computing change over from 64 bit to 128 bit. This takes us up to 2007 + 15 = 2022. Assuming the same basic time changeover occurs, then the move from 128 bit to 256 bit computing could hit around the year 2042. This means the change over might be accelerated to about 2038. Any thoughts as to the problems and issues of the Linus 2038 date issue. It will be less a problem than the W2K damp sqib problem just over 6 years ago. ____________ Account frozen... | |
| ID: 249890 | | |
... But that brings up another problem which is at that time all software, bioses, etc. that depended on the stored date would have to be upgraded exactly at midnight on that date or else it would be reporting the wrong date. As to the y2k bug(feature): With both memory and auxillary storage costing around a buck a byte ($1 for 8 bits) they did an excellant job of saving money. The problem was that most managers went for short term profits and passed the headaches on to their successors instead of implementing changes in a timely manner. If they had started fixing it around 1990 as the programmers planned it would have cost much less. ____________ David Stites Mount Vernon, WA USA  | |
| ID: 249966 | | |
If non volatile ram were to ever become as fast and cheap as standard ram, you could load the "bios", programs, and everything else into it and everything would be fine. Computer bioses could be upgraded the same way you upgrade operating systems today. It wouldn't depend on anything that was hardwired into the system. The only reason todays computers require a "bios" is that the standard ram is wiped clean when it's powered down. ____________ Jim Join GQFCrunchers Problems? Check the Boinc Wiki Or NEW Enhanced FAQ | |
| ID: 249997 | | |
but it's just a program stored on a rewritable ROM chip, so why can't it just be updated like everything else? | |
| ID: 250053 | | |
Well, it can, as far as the bios itself, but I don't know if that would include the data length for the date function or not. Someone with a lot more knowledge of it than me would have to say. But even with a bios upgrade, it's a whole lot more complicated (and dangerous, lost one computer myself that way) to do a bios upgrade than it is to just upgrade software, or even operating systems. So if the contents of the rom bios could be stored at the first data address of a nvram it could be upgraded as easily as overwriting the file in memory. I'm sure the computer would have safeguards so that a program could not inadvertently write to those addresses, but it's the same way with a disk drive now. It would take a specially designed program to write to that address space just like it takes "fdisk" to write to the partition table on a hd. Matter of fact, it takes a special program to update the bios also. However if everything were stored in nvram, and a larger data size were required, all it would do would be to add a few bytes or so to the bios area and remove a few bytes of general memory. With the fixed size of the bios chip this would not be possible. ____________ Jim Join GQFCrunchers Problems? Check the Boinc Wiki Or NEW Enhanced FAQ | |
| ID: 250066 | | |
Not 100% sure of my first fact, but I understand motherboard and BIOS suppliers are looking to move to a new format which does away with the BIOS in the next couple of years. Yep. EFI (not Electronic Fuel Injection) but Extensible Firmware Interface. Can't happen soon enough IMHO. Speculating ... Given the address space offered by 64bit computing i find it hard to see what sort of applications would ever require 128bit computing. Edit- fixed BBCode typos. ____________ Grant Darwin NT. | |
| ID: 250081 | | |
i just find it ignorant of developers/designers to use the easy way (2 digit years for example) which causes many other problems, rather than using a better solution from the start (it's easier to use a different system from the start rather than try to change an existing one) It wasn't that long agao i spent $1,000 for a 32MB SIMM. It wasn't that long ago it cost over $1,000 just for 1MB of RAM. When the decision was made to use 2 digits for the date, it was 50 years away from the turn of the century, and the cost per byte of memory was so high it is hard to comprehend these days. ____________ Grant Darwin NT. | |
| ID: 250086 | | |
Given the address space offered by 64bit computing i find it hard to see what sort of applications would [i]ever[i/] require 128bit computing. Today? I can tell you, point-blank, from working with them: Backend/BackOffice, "industrial-strength" top-shelf database engine servers can certainly use memory access of 64-bit magnitude... I work with SQLServer 2000/2005 daily, & have worked with IBM DB/2 & Oracle 8.x-10x in the past. I.E.-> The more memory they can use, the faster they run, vs. being "diskbound" which today they can be depending on the size of the data... E.G.-> I work in financial & insurance data everyday, & it EASILY crosses the 4gb dataset size mark many times... (64-bit will be a boon to information systems & sciences imo, vs. the 32-bit 4gb per-process addressable memory "limitation" currently nowadays on most systems, of which in Windows @ least, 1/2 (2gb) is used by the Operating System (in std. configurations, it is possible to get 3gb per user process & only 1gb to the OS via the /3gb boot.ini switch) in mgt. of said memory space). APK P.S.=> I'm certain other tasks today (not ones done by normal users, because I have YET to see a 4gb Word.doc made) can benefit from that much memory access/addressability, but that is one I know FOR SURE can gain... apk ____________ http://torry.net/authorsmore.php?id=1781 "The object's hull is made of SOLID neutronium: A single StarShip cannot combat it!" quote Mr. Spock, Star Trek original series, episode title: "The Doomsday Machine" | |
| ID: 250089 | | |
Today? I can tell you, point-blank, from working with them: I am fully aware of that; as i said in my previous post i can't think of any current (or future) software that would need 128bit addressability. 64bits gives 17,179,869,184 gigabytes or 16 exabytes of directly addressable memory- i'm not aware of simulation software that comes anywhere near to even approaching that limit. ____________ Grant Darwin NT. | |
| ID: 250094 | | |
Today? I can tell you, point-blank, from working with them: but software hasn't evolved much compared to software and a few other uses would be in the video and animation fields, for rendering and/or video editing, especially with the increased adoption of HD/HQ video/tv renderers will have to deal with ever larger frames to process, but of course the process will remain mostly CPU and memory bandwidth bound i'm often seeing normal home computers with a gig of ram these days, and quite often it's needed with all the memory hungry apps people want to run at the same time for example if i run firefox (with about 100 tabs), google earth, BOINC, and adobe photoshop, then it'll have used all it's physical ram, anything else will cause it to have to use swap space, so i can see this having quite a few advantages in the home too | |
| ID: 250103 | | |
I am fully aware of that; Cool, this is good that you are, so I can "talk turkey" with you & possibly even get technical, w/out having to form analogies & such when conversing with you on this topical material. as i said in my previous post i can't think of any current (or future) software that would need 128bit addressability. I can - again, for Databases! For the same reasons I gave below - they run much faster with the more memory they can address. Data today, typical data even in industrial environs, probably never hits that mark/size - yet, that is. See, I think that the size of data is subject to this rule typically: IT GROWS! Yes, you can offline store it, but that means loading it again, or the media (probably slower than HDD's are by far) is slower than HDD bound data, and certainly slower than memory based data. Especially for data-warehousing, & views, for something like say, a portal site. I.E. -> If IPv6 ever takes hold, everything will have an IP that is electronic imo, eventually - thus, it's TOTALLY traceable/trackable... To do that, & maintain histories + statistics (limits of your SQL imagination really, lol) well... That means to trace those types of things and much more such as reporting records etc./et all? You'll need TONS of storage... 128-bit tons, eventually, imo. (Tracking/tracing, for GOOD reasons! Not looking @ it here in this example for "sinister" or espionage-type purposes, but literal tracking!) Maybe not THE best example, but here goes an E.G.-> "My presentation data and briefcase/laptop FOR WORK: Damn! Now, where they heck did I put them last night... This sucks, I'm already in a hurry to get to work today on time for that presentation!!" (Go to your PC (or watch probably, lol, by this theoretical time in the future when 128-bit or better is the norm), find out, track the missing/lost item... even if it is @ your girlfriend's place across town!) This is along the lines of what Sam Walton's company "Walmart" is attempting to do with RFID (and I've actually worked with that type of stuff using MSVC++ & a 'Welch-Allyn Dolphin' handheld scanner system for Fedex... it's possible/doable, today, & generates TONS of data... imagine that in tracking CONSTANTLY as to locations when moving/mailing for instance! Systems like it exist today, & I'd wager their storage requirements are hefty, but maintaining history will be staggering) Imagine a time in the future when many if not ALL electronic devices are "hooked into" the net, in near realtime, you'll see my point. 64bits gives 17,179,869,184 gigabytes or 16 exabytes of directly addressable memory- i'm not aware of simulation software that comes anywhere near to even approaching that limit. Right, I'm not either (HUGE numbers)... but, I think that day's coming, & not only because of the "IP Theory" I had above, but also think about voice data, photo/video/streaming video, & just generally BLOB/LOB in nature, & see where I am heading I guess... To the "land of GIANT data" really. :) APK P.S.=> Like I said above, typically, & I have dealt with MIS/IS/IT for around a decade & a 1/2? Data - it just tends to HUGELY grow! apk ____________ http://torry.net/authorsmore.php?id=1781 "The object's hull is made of SOLID neutronium: A single StarShip cannot combat it!" quote Mr. Spock, Star Trek original series, episode title: "The Doomsday Machine" | |
| ID: 250136 | | |
Huh? | |
| ID: 250153 | | |
oops, that was ment to be "but software hasn't evolved much compared to hardware" lol | |
| ID: 250163 | | |
If IPv6 ever takes hold, everything will have an IP that is electronic imo, eventually - thus, it's TOTALLY traceable/trackable... IPv6 is getting there, but a new version of IP isn't the easiest of things to deploy sucessfully (it's gotta be backwards compatible with IPv4 as well) the plans are already in existance, they're finalising the details of the protocol still, but once there's a final spec we'll start seeing it being used (cisco products for example have supported IPv6 for quite a while now) there'll be an address available for every device on the planet (and then some) but that's every device now, the MAC address space is even larger than the IPv6 space (more mac (hardware) addresses than IPv6 addresses), but when that limit is approached, well, things will be very different, because almost nothing of the MAC address space has been used yet, and each address is globally unique (no other device anywhere has the same address) This is along the lines of what Sam Walton's company "Walmart" is attempting to do with RFID (and I've actually worked with that type of stuff using MSVC++ & a 'Welch-Allyn Dolphin' handheld scanner system for Fedex... it's possible/doable, today, & generates TONS of data... imagine that in tracking CONSTANTLY as to locations when moving/mailing for instance!...)RFID is great for situations where security isn't important, because it's easilly hackable, so i wouldn't want it in something like my passport, or credit card Imagine a time in the future when many if not ALL electronic devices are "hooked into" the net, in near realtime, you'll see my point.that'll happen too, and it'll all be in realtime. There's a lot of development going into "all in one" services, the end result being that you have a 100Mbps (1000Mbps eventually) cable/fibre comming into your house/building and all your services run from that, including phone, tv, internet, and other new things that have yet to appear (setting the oven from work?) and this isn't an far off idea, VDSL (the next big thing after ADSL) has speeds of about 52Mbps, so it's entirely possible in the near future however, most internal networks will probably still use IPv4 (because the space is more than big enough for even a massive network) with Network Address Translation to IPv6 for data going onto the net (but each device (computer, handheld, phone, tv, etc.) will have it's own IPv6 addy 64bits gives 17,179,869,184 gigabytes or 16 exabytes of directly addressable memory- i'm not aware of simulation software that comes anywhere near to even approaching that limit.Right, I'm not either (HUGE numbers)... but, I think that day's coming, & not only because of the "IP Theory" I had above, but also think about voice data, photo/video/streaming video, & just generally BLOB/LOB in nature, & see where I am heading I guess... the thing about "potentials" is that it gives possibilities for new technologies and services/ideas to emerge, IPv6 for example allows for some new ways of sending data (instead of one-to-one, one-to-many, one-to-all) there aren't any uses for it at present, but when IPv6 is deployed extensively, i'm sure we'll see things that will use the new methods my point is, new and better ways of doing things can't happen unless the posibilities are there so in data terms, if the current limitations are removed (or at least the limit is lifted a significant ammount) then new things can start to happen you could say the same about 64 bit computing, there aren't any major software packages writen for 64 bit (not natively anyway) to allow them to come into existance, the technology has got to support 64 bit first, then the apps will follow as with everything, it's usually better to have too much, than too little | |
| ID: 250173 | | |
8 bit processors can do 32 bit (or 64 bit) math -- it just takes more instructions for them to do it. ____________ | |
| ID: 250202 | | |
Speaking of memory...I have a complete working install of linux on a USB memory stick...it's bootable. ____________ Account Frozen... | |
| ID: 250212 | | |
Using a 64-bit value introduces a new problem date in about 290 billion years, on Sunday, December 4, 292,277,026,596 15:30:08 UTC. This problem is not, however, widely regarded as a pressing issue. Considering that this system's star will exhaust its fuel, swell out beyond the planet where >99% of the 64-bit systems will be deployed in less than 2% of that interval, I would be highly likely to concur with that assessment... :o) ____________  Stewie: So, is there any tread left on the tires? Or at this point would it be like throwing a hot dog down a hallway? Fox Sunday (US) at 9PM ET/PT | |
| ID: 250363 | | |
Exactly. You've already got most of it there. All you would need was a bios, however with your approach you would still need a bios on the motherboard to be able to access your mem stick. But if you could build something like that into the mobo and add the bios to it, there you go. However your device still requires a battery to keep the memory alive doesn't it? The approach I mentioned in using nvram would not erase itself even if all power were removed. In a normal bios, it still requires a battery to store the date and other configuration information but the code to actually start the boot process is written either to an eeprom or flash rom. Either of which do not erase with no power. Only the stored configuration information, and even then the small onboard battery can keep the info stored for quite a while until it dies. The only problem with storing the entire bios this way is that if the battery ever did die you would have a real problem just to get the computer powered back up! I've got an old IBM Mod 80 386sx that used a special battery that went dead. That was a real pain cause the only thing it would do is access the floppy. I had to put a configuration disk into it to even get it to recognize any hard drives or any other hardware attached, then I still had to go in and change the date and time, etc. It was such a pain to even boot it that I finally stripped it down and junked it! I'd hate to think what kind of trouble it would be if the entire bios were wiped out with each power down! It wouldn't know enough to read a floppy! ____________ Jim Join GQFCrunchers Problems? Check the Boinc Wiki Or NEW Enhanced FAQ | |
| ID: 250374 | | |
I agree...and think it could be done easily. The problem would lie in the cost of the persistant memory. ____________ Account Frozen... | |
| ID: 250376 | | |
Yep, that's the holdup right now. The only other problem I see with this is in case the bios in memory were accidently wiped out how could it be restored without some way of loading it back in. The only solution I can see is to still have a small part of the bios hardwired in with enough instructions to check mem location 0 and if no data was found access another location that had a basic bios that could be restored to allow the computer to work, (say just enough info to be able to access a usb port) where a new copy of the bios could be found and loaded. Or possibly even a "smart card" type system where you get a card with the computer with the bios on it and if a bios upgrade were needed it would be a simple matter of swapping cards. This way if something happened during the upgrade, you could simply reinstall the original card and restore the system. If the bios were copied from the card into nvram, you could remove the card and store it in a safe place while it's not needed. In my first solution above, the tiny amount of code hardwired to the system could say "check for data at address 0, if data found read data and continue boot, if no data found access card reader and copy contents to memory starting at location 0, goto step 2 above". This way nothing about the code stored on the mobo would ever have to be upgraded unless a card reader were changed, and if this is the case the card reader could be shipped with a chip to replace the original code. Or since the access method would more than likely be the same the actual reader control program could be in the reader itself. You could then have a "user space" program for installing a new bios into a running system, sorta like the dos "sys" command. Or else this program could just zero out the bios area and reboot with the new card in the reader. I know that now you can buy upgraded bios chips but you have to actually go inside the computer and swap out chips. The card upgrade could be done by even an unskilled person with no technical knowledge at all. (And yes, I've seen some d***a***es out there that don't even know which end of a screwdriver to use! Sorry if I have hurt anyone's feelings here! haha) ____________ Jim Join GQFCrunchers Problems? Check the Boinc Wiki Or NEW Enhanced FAQ | |
| ID: 250400 | | |
Or possibly even a "smart card" type system... the smart card idea sounds great to me what would be even better is if a few blank ones were shipped with the mo-bo, so the user can "download" a new bios onto one of the blank cards rather than having to order a new card for an upgrade (unless i misinterpreted something, and you said that already) | |
| ID: 250587 | | |
Or possibly even a "smart card" type system... I was thinking about something sorta like that but when I was thinking about it my idea went that maybe the tiny onboard bios could dl the new bios from the manufacturer, but then I thought the code required to find and configure just about any ethernet card (or whatever was in use at the time for a net connection) would be considerably larger than the card idea, plus even though connectivity to the net is increasing daily, what if there were computers at remote sites that need updating when a net connection was not available. But yes, your idea is even better and you would not even have to have the blank cards sent with the computer if they were standardized. You could just go to your local Radio Shack or favorite computer store and buy as many blank ones as you need when you need em. The only keyword here would be "standardized", and how many times have we seen this actually done in computers? (Settling on *just one* "standard"). ____________ Jim Join GQFCrunchers Problems? Check the Boinc Wiki Or NEW Enhanced FAQ | |
| ID: 250598 | | |
Program developers will continue using it and start to panic in 2037 nah the same thing will happen with it as PC's with Y2K. the real problem will be solved many years before the press even heard about it, due to things like banks need a large future date predion buffers. Then in 2038 the Press will kick up a big drama fuss when some points out to them that *nix is used for backbone stuff and that if they break everythign breaks, and pick up on the search for the lost few remain non mission critcal legasy search, and once agians fail to realise that the important stuff was fixed years before. ____________ uninstall dyslexica.o : Permission denied  AMD Athlon 64 3000+ w/Windows AMD Athlon 1800+ w/Linux | |
| ID: 250602 | | |
true, but that might be a problem for those who live in rural areas but have net accesswhat would be even better is if a few blank ones were shipped with the mo-bo, so the user can "download" a new bios onto one of the blank cards rather than having to order a new card for an upgrade... The only keyword here would be "standardized", and how many times have we seen this actually done in computers? (Settling on *just one* "standard").indeed, not often, hardware is the best example and even when there are standards, they aren't always followed, look at web standards lol most sites don't use valid code, and a lot of "web masters" don't know what these standards are (i've met a few who actually didn't know lol, and in another thread about web standards, someone said that in an email he got from the company that did the SETI site design, when asked about why they didn't write standards compliant code, replied with "what are these "standards" of which you speak?" oh dear!) | |
| ID: 250610 | | |
|
Well problem With Web is When a lot of the webmast learn to Code, the IE/ Netscape tag battle was in full swing. so ther was not "apparent standard, even some of those that know they existed wrote "Invalid code" simply so their sites would work on both of the leading browers. and got into a habit uf uses tag soup that woork and forgetting to make it as close to the standard as possible.... | |
| ID: 250670 | | |
|
Ok, first, if you lived in a rural area with net access, using the smart card idea, how many people nowadays lives so far from some type of electronics or computer store that they couldn't make a trip there at least sometime to pick up a few blank cards? And if not, you have net access so order em on the net and have em mailed to you. And yes, smart cards would be fairly cheap, especially to a big computer manufacturer, so send at least one blank one with the computer. | |
| ID: 250696 | | |
Well problem With Web is When a lot of the webmast learn to Code, the IE/ Netscape tag battle was in full swing. so ther was not "apparent standard, even some of those that know they existed wrote "Invalid code" simply so their sites would work on both of the leading browers. and got into a habit uf uses tag soup that woork and forgetting to make it as close to the standard as possible....well that's all because the older browsers are broken, well, IE is still broken, propritry tags should never be used, because that instantly breaks the ideals of the web (viewable in any browser) but i agree that to get a site to work, especially in more than one browser, required a lot of code hacks, but there's no excuse for it these days, and the general opinion is that "professional" web designers/webmasters who don't know how to do things properly, can't call themselves professionals, for the simple reason that they don't know their craft very well if they don't know about, and adhere to standards what would you think of a locksmith who didn't use standard shape/size blank keys, and made his own? I'd rather have the standard key myself, because it's more likly to work in my lock than a custom shaped one (i'm talking about the general key design, not how the teeth are cut) | |
| ID: 250934 | | |
Ok, first, if you lived in a rural area with net access, using the smart card idea, how many people nowadays lives so far from some type of electronics or computer store that they couldn't make a trip there at least sometime to pick up a few blank cards? And if not, you have net access so order em on the net and have em mailed to you. And yes, smart cards would be fairly cheap, especially to a big computer manufacturer, so send at least one blank one with the computer.true, to be honest i don't know, where i live everything is pretty close together, i'm in what most people would call the "country" but the "town/city" is only a 20 minute drive away (with a 40 mph speed limit, so not that far) And standards, yes I agree, but I think the manufacturers and aftermarket suppliers could get together and decide on at least a couple of different standards that all the manufacturers could use, say one for IBM compatibles, one for Macs, and maybe one or two others for non-mainstream computers. (I'm not saying that to belittle any computer that is not an IBM compat or Mac, but just that there are more of those two in general use.) And to think of it, it would be a sight to see if all the computer manufacturers would sit down together and develop a card system that any computer could use interchangeably! That would be a historical first!indeed, and i imagine that there would be a few different ones for different things (like there are different image formats, and case/motherboard standards) to cater for the differing requirements, the main thing is that each be an open standard, not proprietry (sp?) so that everyone can follow the same standards and not have to shell out loads of cash to another company, which just results in each company making it's own standard, which doesn't help anyone look at the flash card industry, there are quite a few types, compact flash, SD, and quite a few others that i can't remember, and all in all it works kinda well. Probably not the best example because as far as i can see there isn't much need for different formats, because they all do the same sort of job(s) i think it would be best for a standards body to decide what standard is best, so that each company isn't trying to have things their way, and decisions aren't influenced by money (having something reliable is better for everyone than something that's cheap) The "internet html standards", yep I was right in the middle of it....all i'm gonna say is "oh dear", that must have been erm, "fun" lol I only started getting into web dev about a year ago, so i'm learning the new ways, which i'm quite glad about really wouldn't have liked to be around when M$ and netscape were at playing tug of war (see, when companies are left to decide standards they don't do a good job) | |
| ID: 250936 | | |
|
Okay, it's time for a little bit of a reality check, folks. | |
| ID: 250941 | | |
...on PC machines, we don't have to worry about the BIOS on anything reasonably modern.... so this is a software only problem? cool :) i still like the smart-card idea for bios upgrades, it just seems easier and safer (imagine a power cut while you're flashing your bios ;) lol) | |
| ID: 250951 | | |
well that's all because the older browsers are broken, well, IE is still broken, propritry tags should never be used, because that instantly breaks the ideals of the web (viewable in any browser) Hrm..... As I see it, it isn't so much proprietary tags as differences in implementation. A simple HTML table should be something like: <table> <tr><td>1a</td><td>1b</td></tr> <tr><td>2a</td><td>2b</td></tr> </table> Now, if I'm trying to write something that goes really fast, I can not process (skip over) most of the closing tags, and I still know everything I need to displya the table correctly -- an opening tag implies a closing tag. If I choose to implement my web browser that way, it will correctly display the table above according to all standards, and it will also correctly display this table: <table> <tr><td>1a<td>1b <tr><td>2a<td>2b </table> So, is this a "proprietary tag" or simply malformed HTML that renders OK because it takes advantage of how the browser is actually implemented? There are lots of places like this where the browser either takes an optimization that permits malformed code to work, or where the implemetors have simply read the standard and interpreted it differently. ____________ | |
| ID: 250952 | | |
...on PC machines, we don't have to worry about the BIOS on anything reasonably modern.... Remember: the nice thing about standards is that there are so many to choose from. For most of us, the BIOS is just a bootstrap: it provides enough software (in ROM -- not Mr. Walton, the other kind of ROM) to read the master boot record off of the hard disk and start executing it. Unless you need some BIOS "feature" that is part of an upgrade (like support for a new CPU that didn't exist with the motherboard was made) there is rarely a good reason to upgrade. BIOSes used to come on physical media -- you ordered the BIOS upgrade and it came on EPROMs. Many new motherboards have "dual bios" chips, so if something fails during the write, you've got another BIOS, just in case. ____________ | |
| ID: 250954 | | |
Um you better check the date on that RTC revtion your holding to 8) it was not that the the 386 and old did not bother to read the up two, it was not there then IF it was there a BIOS update would have been possible on some of the 386 to have th correct code, thyu could no do that becuase at the time the RTC only asked for MM/DD/YY ____________ uninstall dyslexica.o : Permission denied AMD Athlon 64 3000+ w/Windows AMD Athlon 1800+ w/Linux | |
| ID: 250964 | | |
Actually, the RTC doesn't "ask" at all. It is the BIOS that talks to the clock. But, all of that notwithstanding, there are at least two really simple solutions. The easiest is to get the two digit year, and if it's "80" or greater, then the century value is "19" -- if it's less than 80, then the century is "20" and we're good until 2080. The next one is to pick any unused CMOS memory location and stuff the century there. For the "bonus round" the OS can simply ignore the century as supplied by the BIOS/RTC and as long as it knows the correct time within a century or so we can still get the full, correct year. (knowing that time moves forward is a useful thing here -- just storing the date on the HD once in a while will do nicely) ____________ | |
| ID: 250971 | | |
true, implementation has a lot to do with it, and that's why things are moving towards strict languages like XML and XHTML where you can't "bend the rules" it's either correct (and working) or not, which personally is the correct way to do it, assumptions in the web arena are generally never a good thing (take character encoding for example, and all the problems with that)well that's all because the older browsers are broken, well, IE is still broken, propritry tags should never be used, because that instantly breaks the ideals of the web (viewable in any browser)As I see it, it isn't so much proprietary tags as differences in implementation. i was just making the point about propritry tags | |
| ID: 250976 | | |
Unless you need some BIOS "feature" that is part of an upgrade (like support for a new CPU that didn't exist with the motherboard was made) there is rarely a good reason to upgrade. what about bug fixes or better ways of doing things (better coding) etc. i know that in an ideal world these wouldn't be needed, but we don't live in an ideal world | |
| ID: 250979 | | |
Unless you need some BIOS "feature" that is part of an upgrade (like support for a new CPU that didn't exist with the motherboard was made) there is rarely a good reason to upgrade. I don't think any of the reasonably modern operating systems use BIOS services extensively -- if that's true, the only thing we're really talking about here is initialization (i.e recognizing a "Pentium-M" vs. "Pentium-4") or dealing with hard drives with larger partitions (like happend a while back when we passed 137gb). Who cares if "better coding" saves a couple of seconds during boot, and then is bypassed. ____________ | |
| ID: 250989 | | |
Unless you need some BIOS "feature" that is part of an upgrade (like support for a new CPU that didn't exist with the motherboard was made) there is rarely a good reason to upgrade. While it's true that most everything could be left alone and not bother with an upgrade but new technology comes out all the time. What about USB? This wasn't out too long ago and even though you could get cards to add usb ports to your computer, if you were using a usb mouse or keyboard and had to access the setup program to change the configuration you would have to dig out a standard keyboard to do it with since the bios on the older computer did not recognize the usb ports. So if you have some new technology like that to come out in the future, wouldn't you like to be able to just add a card and keep using your computer, or are you going to have to buy a new computer to take advantage of the new technology. The smart card would allow you to simply add an adapter card to the computer and allow it to be used to it's fullest without replacing the whole computer or not being able to use the technology 100% of the time. For instance a new disk drive technology. Suppose someone came out with a terrabyte drive that was as fast as ram but used a totally different address method. And you wanted to boot from this drive. Two choices, junk your present computer or upgrade the bios, and the easiest and safest way to do that would be with something similar to the smart card. IMHO ____________ Jim Join GQFCrunchers Problems? Check the Boinc Wiki Or NEW Enhanced FAQ | |
| ID: 252102 | | |
Unless you need some BIOS "feature" that is part of an upgrade (like support for a new CPU that didn't exist with the motherboard was made) there is rarely a good reason to upgrade. POST is a little bit more involved than that. ____________ Account Frozen... | |
| ID: 252103 | | |
Using a 64-bit value introduces a new problem date in about 290 billion years, on Sunday, December 4, 292,277,026,596 15:30:08 UTC. This problem is not, however, widely regarded as a pressing issue. OK, that's it. I'm recoding all date functions to use BigDecimal instead of a measly int. Integer part of the BigDecimal will be the number of seconds since Unix Epoch (just to keep things compatible :D). Decimal part will be the fractional number of seconds. Benefits: * Easily allows for the expression of any time, be it past, present, or future. * Integer part directly compatible with Unix Epoch. * Decimal part can be made as precicely as one wishes (for example, if measuring less than a planck time ever turns out to be important :D) Problems: * Can you say "memory hog"? * Slower than integer to update * Arbitrarily sized (making it harder to read/parse dates from files) * Fewer libraries support BigDecimal than support integers or 64-bit integers. (...ok, so I'm not actually. "64-bit should be enough for anybody!" ;)) Puffy ____________  | |
| ID: 252358 | | |
Problems:well, another problem is that 64-bit isn't as efficient with some thing, like when representing small values, or just plain old 0 (zero) having all those extra digits just slows things down, more data to shift around etc. so what's needed is an extensible format really, best of both worlds then, but of course backwards compatibility is going to be an issue with a new format ____________ Want to search the BOINC Wiki, BOINCstats, or various BOINC forums from within firefox? Try the BOINC related Firefox Search Engines | |
| ID: 252364 | | |
Message boards : Number crunching : Year 2038 Problem (Devs?)
| Copyright © 2009 University of California |
