there is an easy soloution to that problem on linux, chroot. The outer system can contain an up to date kernel and all the hardware drivers needed while the chroot can contain any version of the standard libs you desire. The linux syscall interface is one of the most backwards compatible interfaces arround.
on windows i find its rare that apps refuse to work on a new version but maybe some itnerfaces are worse than others in this regard.
I suggest that you take a look at basic utilities like tar,/bin/sh, and man in the UNIX or Linux world. thats why you code in a proper programming language using proper stable interfaces, shell scripting (or using the shell commands from a proper programming language) is a dirty hack that will bite you later.
what serious breaking changes have there been in the "core os and apis"?
afaict it was bad from the start, iirc military requirements gave it the insane plane like design (to allow it to land covertly) and the larger than nessacery size. The benifits of reusability were never really realised due to insane refit costs and timespans. Sure it can do some cool shit like bringing home LEO satalites but its very rare that there is a real need to do that.
trouble is the ISS has locked america in to keeping the shuttle program running in the medium term (plans for its long term death are already made). All the modules (many of which are from partner countries making just scrapping them or forcing major rework politically disasterous) are built to go up in a shuttle cargo bay.
there will be a time for reusable spacecraft but that time will not come until we move on from barely adequate chemical rockets to nuclear power.
MS came up with the idea of CD autorun in the days before CD burners so virus spread that way wasn't really feasible and presumablly they thought that those with the rescources to produce CDs would be professional enough not to do anything more than load a menu system or similar.
CDs with both music and data are nothing new and record companies have been using them for videos for some time. They rely on the fact that audio players read the first session while CD rom drives (at least modern ones) look for the data index structures in the last session.
many savvy users turn CD autorun off but its pretty much inevitable that sooner or later even they will leave a machine on the defaults by mistake.
i'm pretty sure on both windows and linux a normal user can make stuff run when they login pretty easilly and on many linux installs they can probablly even use cron to schedule it to happen soon after startup.
IIRC, Windows 2000 Ceritified Programs were supposed to remedy this by being need-to-install-as-admin but can-be-run-by-anyone by design but of course what proportion of software vendors actually enter for said certification programs?
to the "Slocket" fiasco what exactly was the fiasco regarding slockets (devices for putting socket 370 processors in slot 1 motherboards) and why did slockets never appear for AMD processors?
slots were an abberation that both AMD and intel jumped on and then soon dropped and afaict AMD and intel both did it for the same reasons, coupling the L2 cache though a CPU socket wasn't really workable at the speeds amd and intel wanted and in package L2 cache drove yeilds though the floor (remember the pentium pro?). Both companies then moved to on die L2 cache making the slot based package redundant.
your sequence has a couple of problems, a fab is only valuable for PC CPU production for a relatively short time. Just as PCs devalue very quickly due to the constant advances in the industry so does the equipment used to make them. There is also the reuputation issue, remember the old saying "nobody ever got fired for buying IBM"? if you buy from the dominant market force and it fails your boss will likely put it down to bad luck. if you buy from a new upstart and it fails your job is on the line. Also what really happens to company A's assets is they are sold (possiblly at far less than thier production costs) to a company that can make use of them for some other use.
Patents are also a major barrier to entry in the industry (though you may rightly consider that goverment medling)
look at cyrix, they made a mistake during one generation (poor FPU in an otherwise very fast pentium class CPU just as software really started using floating point intensively) and were permanently relegated to the low end market thier carcus being split between via (who make low speed low power PC compatible chips aimed at embedded and thin client markets) and amd (intels only significant current competitor).
But anyway, the IRC network model is a pretty neat one not really, its a very old sucky design with far too much server to server replication and trust making it suicidal to run an open irc network (ever heared of eris?).
lots of local servers linked together to form networks lots of closed networks that don't interoperate with each other.
very few centralized points of failure every link is a major point of failure splitting the network in two.
direct connections for file transfer rather than pushing them through the network most im networks only push filetransfers through the network if direct connections fail.
I think it's too bad in some ways that IM arose from centralized models that lend themselves to corporate fiefdoms. agreed but don't think irc is any better in this regard small networks are often run by people with big egos who love to bully thier users. Large networks are run by people who've given up on caring about thier users (tried to report a troublemaker on undernet or quakenet recently?).
I find Jabber pretty neat because of the ways you can link servers I find jabber neat because you don't have to explicitly link servers or trust the servers you link to. Links are just made automatically as your users require them.
and communicate from one server to another -- perhaps less like IRC than email far less like irc and far more like e-mail. IRC is closed networks running on open protocols (at least an open client side protocol, the server to server protocols are often custom and poorly documented). E-mail and jabber are open systems running on open protocols.
but I think it'll be a long time before we see the demise of the big AIM/MSN/Yahoo/ICQ networks, even if at some point in the far future they're seen as nothing but a quirk of the early development of the Internet. sadly i belive you are right.
its great for chatting with a room full of people and occasionally private messaging those same people but support for contact lists on irc is abysmal:
1: there is no widely supported way to watch a user for changes other than polling and poll frequency is very limited by ircs built in abuse limits especailly if you wan't to monitor the hostnames and away status. 2: I don't know of any major irc network that really enforces nick ownership properly (read: nick passwords checked at login time with no way to use a nick (even briefly) without the password. Some major networks don't even have nickserv at all. 3: IRC is based on relatively small closed networks, no network has either the size of MSN/AIM/ICQ/YIM or the open nature of jabber.
I can understand that Windows 3.x and Windows 9x, with all their DOS tradition behind them had to be "admin-by-default". it wasn't just admin by default it was zero security infrastructure at all. Once a binary was running it could do whatever it liked the only security features were just UI lockdowns.
But since Microsoft moved to a new architecture, why don't they change the default user profile behaviour ? Old APPs are run thrue an emulated API i belive this is what they plan to do in vista
newer application break if they can't run in a non-priviledged environnement. can't really be done since from a binary point of view there is no real way to tell the difference between an app from the 9x days and one written yesterday. The fake admin sandbox will have to availible for for use with any app or not availible at all.
afaict there are two main MS license programms that are specifically for academic institutions. I'm not sure what the prices are but i suspect its a relatively small portion of the tuition fees.
Campus agreement makes MS software (OS upgrade/downgrade, office, visual studio and various other stuff) availible for installation on any university machine with no need to count the number of installs (though the evidence of the OEM licenses for windows must be maintained and if a machine doesn't have an OEM license its time to pay retail or find a vendor that breaks its OEM agreement). It destroys any cost arguments against using MS on a particular machine by taking a subscription which is not tied to the number of installs (i think its based on total student count or something like that but i'm not sure). It also brings a huge degree of lockin as an institution withdrawing from the programm would need to relicense or remove any software installed under it.
MSDN academic advantage provides departments (the EEE department i'm in isn't signed up so i'm not sure on the details) with free/cheap software for student use on both university machines and thier own machines and includes stuff like the server versions of windows, i belive this is the source of those very cheap copies of MS software you mentioned. However software obtained under this license cannot be used by the department for any purpose other than teaching use of the software in question.
Everyone I have showed Gaim to have not only switched, but suggested it to their friends i'm running gaim at the moment for jabber support (not having much luck pushing jabber to my friends though much as i'd like to reduce dependence on the big 4 im networks that have a history of hostility towards multiprotocol clients) but its interface feels far more clunky than trillian and basic features like a tray icon are hidden in the plugins section.
If theese people are currently unaware of multiprotocol clients at all then i could see gaims attractiveness but i'd be surprised if you could get trillian users to switch so easilly.
last i checked balloon gas was a helium/air mixture for both safety (people are going to inhale the contents of baloons to change thier voices pitch, you really don't wan't doing so to be suicidal) and cost (helium is expensive) reasons.
apart from a few true classics, people don't tend to buy old movies or go to the cinema to watch them, they watch them on TV (a channel for which the gaming world has no real equivilent), the bulk of the market is producing mediocre movies using the latest technology and those movies really do date. We just remember the good ones.
As for the computer market many classics (mario, sonic, doom, duke and quake to name a few) are perserved in a legal mannor and many more are preserved though illegal means (pirate collections of rom images etc). However one BIG problem is that moving games to a new platform is far more expensive than doing the same with film/video. Emulation is really difficult to do well for consoles that have 3D acceleration. Porting is a lot of effort and many companies don't wan't to release the source (and on the console side you have the control freak console vendors in the way as well).
And with this no need to plan expensive real terrorist acts, you just have to convince the USA intelligences services that something big is going to happen. not really, if a while passes with lots of alerts and no actual terrorism people will just start ignoring the alerts. Keeping the public terrified long term is going to require actual attacks.
with older versions of linux distros and all versions of windows installing vmware tools is pretty much essential to get the graphics driver (with modern distros this isn't an issue as vmware put the driver under an appropriate license for inclusion in Xfree86).
other than that the clock sync is handy especially if you use the suspend feature (which has the side effect of freezing the guests clock for the duration of the suspend leaving it way out of whack when you resume) and so is the ability to move your mouse over the edge of a VM window to get in and out.
prohibition has the effect of putting all suppliers of a substance outside the normal legal framework. Buying anything from unlicensed street vendors is a risky buisness as they can sell you fake and then dissapear (shop owners and other licensed traders can do this as well to some extent but for them starting anew has a much greater cost possiblly involving fake ID).
imo the only sane approach is to limit regulation to enforcing accurate labeling (e.g if it says its cannibis it had better be cannibis at the strength specified), maybe have a list of standard filler ingrediants that don't need to be declared to stop the labeling requirements getting silly.
i just don't see how pushing money in the direction of gangs and destroying the ability to regulate the quality control of stuff people ingest is a good idea.
i seem to remember reading they don't replace the failed ones, they just junk the entire rack when it becomes not worth running anymore.
i'm guessing google are big enough to have thier own datacenters and thus not have space at such a premium as smaller operations. If space isn't at a premium replacing a machine in a rack probablly isn't worth it (it means you have a machine whose remaining usefull life is out of sync with the rest of the rack its in).
i'm english so i don't really care about case rules for foriegn languages but i can see how they could be a pain, i'm generally of the school of thought that says filenames should be restricted to ASCII.
the fact that many users of case sensitive systems resort to all lowercase to avoid the problems of case sensitivity tell me that its usability effect is anything but slight. You try to note down a URL with significant mixed case (say something interesting you found while using a public terminal to fill in time or something a friend has just shown you), unless your handwriting is very neat and you have a good writing surface its going to be tricky.
of course whats worse than either case sensitivity or case insensitivity is having to deal with both. Your code builds fine on windows but when you go to compile it on linux you find the case is all wrong, similarlly (though nowhere near as common) your code builds fine on linux but when you try and extract it on a windows box files extract over the top of each other.
as for programming languages i think that has more to do with the dominance of C than with the actual pros and cons of case sensitivity.
1. the requirement the ".3" portion be satisfied, i.e., if you didn't give a ".3" extension, it wasn't valid. untrue you could have files without an extention under plain dos.
2. the semantic mapping of the extension to filetype, WTF? so they made the filetype part of the name so you could have identically named files of different types (e.g. the source and the binary of an app), seems sensible enough to me.
3. the implied (don't remember if it was canonical) semantic that no ".3" extension meant the file was a directory directories were identified by a seperate attribute, as i said above you could have an extentionless file, you could also have a directory with an extention (though it was rare)
4. the case insensitive nature of file names very much a personal preference thing this.
they could have put in (and heavilly pushed the use of) an api to allow backup apps to backup and restore short filenames correctly though.
i guess they could have also used a random number rather than sequential, they couldn't really use an alias scheme that didn't contain any of the original filename as it would have made life near impossible for users of old apps.
Slashdot Mirror
there is an easy soloution to that problem on linux, chroot. The outer system can contain an up to date kernel and all the hardware drivers needed while the chroot can contain any version of the standard libs you desire. The linux syscall interface is one of the most backwards compatible interfaces arround.
on windows i find its rare that apps refuse to work on a new version but maybe some itnerfaces are worse than others in this regard.
I suggest that you take a look at basic utilities like tar, /bin/sh, and man in the UNIX or Linux world.
thats why you code in a proper programming language using proper stable interfaces, shell scripting (or using the shell commands from a proper programming language) is a dirty hack that will bite you later.
what serious breaking changes have there been in the "core os and apis"?
afaict it was bad from the start, iirc military requirements gave it the insane plane like design (to allow it to land covertly) and the larger than nessacery size. The benifits of reusability were never really realised due to insane refit costs and timespans. Sure it can do some cool shit like bringing home LEO satalites but its very rare that there is a real need to do that.
trouble is the ISS has locked america in to keeping the shuttle program running in the medium term (plans for its long term death are already made). All the modules (many of which are from partner countries making just scrapping them or forcing major rework politically disasterous) are built to go up in a shuttle cargo bay.
there will be a time for reusable spacecraft but that time will not come until we move on from barely adequate chemical rockets to nuclear power.
MS came up with the idea of CD autorun in the days before CD burners so virus spread that way wasn't really feasible and presumablly they thought that those with the rescources to produce CDs would be professional enough not to do anything more than load a menu system or similar.
CDs with both music and data are nothing new and record companies have been using them for videos for some time. They rely on the fact that audio players read the first session while CD rom drives (at least modern ones) look for the data index structures in the last session.
many savvy users turn CD autorun off but its pretty much inevitable that sooner or later even they will leave a machine on the defaults by mistake.
i'm pretty sure on both windows and linux a normal user can make stuff run when they login pretty easilly and on many linux installs they can probablly even use cron to schedule it to happen soon after startup.
IIRC, Windows 2000 Ceritified Programs were supposed to remedy this by being need-to-install-as-admin but can-be-run-by-anyone by design
but of course what proportion of software vendors actually enter for said certification programs?
to the "Slocket" fiasco
what exactly was the fiasco regarding slockets (devices for putting socket 370 processors in slot 1 motherboards) and why did slockets never appear for AMD processors?
slots were an abberation that both AMD and intel jumped on and then soon dropped and afaict AMD and intel both did it for the same reasons, coupling the L2 cache though a CPU socket wasn't really workable at the speeds amd and intel wanted and in package L2 cache drove yeilds though the floor (remember the pentium pro?). Both companies then moved to on die L2 cache making the slot based package redundant.
your sequence has a couple of problems, a fab is only valuable for PC CPU production for a relatively short time. Just as PCs devalue very quickly due to the constant advances in the industry so does the equipment used to make them. There is also the reuputation issue, remember the old saying "nobody ever got fired for buying IBM"? if you buy from the dominant market force and it fails your boss will likely put it down to bad luck. if you buy from a new upstart and it fails your job is on the line. Also what really happens to company A's assets is they are sold (possiblly at far less than thier production costs) to a company that can make use of them for some other use.
Patents are also a major barrier to entry in the industry (though you may rightly consider that goverment medling)
look at cyrix, they made a mistake during one generation (poor FPU in an otherwise very fast pentium class CPU just as software really started using floating point intensively) and were permanently relegated to the low end market thier carcus being split between via (who make low speed low power PC compatible chips aimed at embedded and thin client markets) and amd (intels only significant current competitor).
But anyway, the IRC network model is a pretty neat one
not really, its a very old sucky design with far too much server to server replication and trust making it suicidal to run an open irc network (ever heared of eris?).
lots of local servers linked together to form networks
lots of closed networks that don't interoperate with each other.
very few centralized points of failure
every link is a major point of failure splitting the network in two.
direct connections for file transfer rather than pushing them through the network
most im networks only push filetransfers through the network if direct connections fail.
I think it's too bad in some ways that IM arose from centralized models that lend themselves to corporate fiefdoms.
agreed but don't think irc is any better in this regard small networks are often run by people with big egos who love to bully thier users. Large networks are run by people who've given up on caring about thier users (tried to report a troublemaker on undernet or quakenet recently?).
I find Jabber pretty neat because of the ways you can link servers
I find jabber neat because you don't have to explicitly link servers or trust the servers you link to. Links are just made automatically as your users require them.
and communicate from one server to another -- perhaps less like IRC than email
far less like irc and far more like e-mail. IRC is closed networks running on open protocols (at least an open client side protocol, the server to server protocols are often custom and poorly documented). E-mail and jabber are open systems running on open protocols.
but I think it'll be a long time before we see the demise of the big AIM/MSN/Yahoo/ICQ networks, even if at some point in the far future they're seen as nothing but a quirk of the early development of the Internet.
sadly i belive you are right.
its great for chatting with a room full of people and occasionally private messaging those same people but support for contact lists on irc is abysmal:
1: there is no widely supported way to watch a user for changes other than polling and poll frequency is very limited by ircs built in abuse limits especailly if you wan't to monitor the hostnames and away status.
2: I don't know of any major irc network that really enforces nick ownership properly (read: nick passwords checked at login time with no way to use a nick (even briefly) without the password. Some major networks don't even have nickserv at all.
3: IRC is based on relatively small closed networks, no network has either the size of MSN/AIM/ICQ/YIM or the open nature of jabber.
I can understand that Windows 3.x and Windows 9x, with all their DOS tradition behind them had to be "admin-by-default".
it wasn't just admin by default it was zero security infrastructure at all. Once a binary was running it could do whatever it liked the only security features were just UI lockdowns.
But since Microsoft moved to a new architecture, why don't they change the default user profile behaviour ? Old APPs are run thrue an emulated API
i belive this is what they plan to do in vista
newer application break if they can't run in a non-priviledged environnement.
can't really be done since from a binary point of view there is no real way to tell the difference between an app from the 9x days and one written yesterday. The fake admin sandbox will have to availible for for use with any app or not availible at all.
I would like people talking to me to be using Gaim so I can use gaim-encryption or OTR (via gaim-otr) to encrypt conversations
just how secure are OTR and gaim-encryption anyway? in particular how do they handle key exchanges (e.g. what is there to stop a MITM attack?)
and does it work with third party clients or just the official ones?
afaict there are two main MS license programms that are specifically for academic institutions. I'm not sure what the prices are but i suspect its a relatively small portion of the tuition fees.
Campus agreement makes MS software (OS upgrade/downgrade, office, visual studio and various other stuff) availible for installation on any university machine with no need to count the number of installs (though the evidence of the OEM licenses for windows must be maintained and if a machine doesn't have an OEM license its time to pay retail or find a vendor that breaks its OEM agreement). It destroys any cost arguments against using MS on a particular machine by taking a subscription which is not tied to the number of installs (i think its based on total student count or something like that but i'm not sure). It also brings a huge degree of lockin as an institution withdrawing from the programm would need to relicense or remove any software installed under it.
MSDN academic advantage provides departments (the EEE department i'm in isn't signed up so i'm not sure on the details) with free/cheap software for student use on both university machines and thier own machines and includes stuff like the server versions of windows, i belive this is the source of those very cheap copies of MS software you mentioned. However software obtained under this license cannot be used by the department for any purpose other than teaching use of the software in question.
Everyone I have showed Gaim to have not only switched, but suggested it to their friends
i'm running gaim at the moment for jabber support (not having much luck pushing jabber to my friends though much as i'd like to reduce dependence on the big 4 im networks that have a history of hostility towards multiprotocol clients) but its interface feels far more clunky than trillian and basic features like a tray icon are hidden in the plugins section.
If theese people are currently unaware of multiprotocol clients at all then i could see gaims attractiveness but i'd be surprised if you could get trillian users to switch so easilly.
last i checked balloon gas was a helium/air mixture for both safety (people are going to inhale the contents of baloons to change thier voices pitch, you really don't wan't doing so to be suicidal) and cost (helium is expensive) reasons.
apart from a few true classics, people don't tend to buy old movies or go to the cinema to watch them, they watch them on TV (a channel for which the gaming world has no real equivilent), the bulk of the market is producing mediocre movies using the latest technology and those movies really do date. We just remember the good ones.
As for the computer market many classics (mario, sonic, doom, duke and quake to name a few) are perserved in a legal mannor and many more are preserved though illegal means (pirate collections of rom images etc). However one BIG problem is that moving games to a new platform is far more expensive than doing the same with film/video. Emulation is really difficult to do well for consoles that have 3D acceleration. Porting is a lot of effort and many companies don't wan't to release the source (and on the console side you have the control freak console vendors in the way as well).
And with this no need to plan expensive real terrorist acts, you just have to convince the USA intelligences services that something big is going to happen.
not really, if a while passes with lots of alerts and no actual terrorism people will just start ignoring the alerts. Keeping the public terrified long term is going to require actual attacks.
with older versions of linux distros and all versions of windows installing vmware tools is pretty much essential to get the graphics driver (with modern distros this isn't an issue as vmware put the driver under an appropriate license for inclusion in Xfree86).
other than that the clock sync is handy especially if you use the suspend feature (which has the side effect of freezing the guests clock for the duration of the suspend leaving it way out of whack when you resume) and so is the ability to move your mouse over the edge of a VM window to get in and out.
prohibition has the effect of putting all suppliers of a substance outside the normal legal framework. Buying anything from unlicensed street vendors is a risky buisness as they can sell you fake and then dissapear (shop owners and other licensed traders can do this as well to some extent but for them starting anew has a much greater cost possiblly involving fake ID).
imo the only sane approach is to limit regulation to enforcing accurate labeling (e.g if it says its cannibis it had better be cannibis at the strength specified), maybe have a list of standard filler ingrediants that don't need to be declared to stop the labeling requirements getting silly.
i just don't see how pushing money in the direction of gangs and destroying the ability to regulate the quality control of stuff people ingest is a good idea.
i seem to remember reading they don't replace the failed ones, they just junk the entire rack when it becomes not worth running anymore.
i'm guessing google are big enough to have thier own datacenters and thus not have space at such a premium as smaller operations. If space isn't at a premium replacing a machine in a rack probablly isn't worth it (it means you have a machine whose remaining usefull life is out of sync with the rest of the rack its in).
i'm english so i don't really care about case rules for foriegn languages but i can see how they could be a pain, i'm generally of the school of thought that says filenames should be restricted to ASCII.
the fact that many users of case sensitive systems resort to all lowercase to avoid the problems of case sensitivity tell me that its usability effect is anything but slight. You try to note down a URL with significant mixed case (say something interesting you found while using a public terminal to fill in time or something a friend has just shown you), unless your handwriting is very neat and you have a good writing surface its going to be tricky.
of course whats worse than either case sensitivity or case insensitivity is having to deal with both. Your code builds fine on windows but when you go to compile it on linux you find the case is all wrong, similarlly (though nowhere near as common) your code builds fine on linux but when you try and extract it on a windows box files extract over the top of each other.
as for programming languages i think that has more to do with the dominance of C than with the actual pros and cons of case sensitivity.
The flags were hidden, but they manifested in the icon.
i seem to remember macs like windows let executables specify thier own icons.
most lusers use the default icons for a file type.
combine the two and you can easilly sucker in lusers to running your app.
1. the requirement the ".3" portion be satisfied, i.e., if you didn't give a ".3" extension, it wasn't valid.
untrue you could have files without an extention under plain dos.
2. the semantic mapping of the extension to filetype, WTF?
so they made the filetype part of the name so you could have identically named files of different types (e.g. the source and the binary of an app), seems sensible enough to me.
3. the implied (don't remember if it was canonical) semantic that no ".3" extension meant the file was a directory
directories were identified by a seperate attribute, as i said above you could have an extentionless file, you could also have a directory with an extention (though it was rare)
4. the case insensitive nature of file names
very much a personal preference thing this.
they could have put in (and heavilly pushed the use of) an api to allow backup apps to backup and restore short filenames correctly though.
i guess they could have also used a random number rather than sequential, they couldn't really use an alias scheme that didn't contain any of the original filename as it would have made life near impossible for users of old apps.