do you know of anything that was broken besides SBCL (which was how this was discovered in the first place).
BTW I belive the intention of debian is to attack this problem from all sides. Afaict SBCL is being changed to keep the direction flag set for as short a time as possible. gcc is being changed to return to the older less likely to fail behaviour and linux is being changed to do what it should have done in the first place.
Well afaict the debian developers plan to modify gcc 4.3 so it behaves in the old way to reduce the risk of crashes when upgrading from one version of debian to the next. Dunno if gcc upstream will agree on that reasoning though. This isn't perfect though, even before gcc's behaviour changed there was still a risk that a signal handler would break the code that it interrupted.
Afaict this bug only affects a relatively small number of apps because little code messes with the direction flag in the first place
VHDL and verilog are a bit weired in the sense that they were originally designed as a method of describing hardware so it could be simulated and were later (ab)used for synthisis. This means that there are a lot of constructs that are perfectly legal verilog but will either cause an error in the synthisis tool or will synthisize to give different behaviour from thier simulation beaviour.
An FPGA is more like a vast sea (or field) of disconnected NAND NOR gates. maybe some old FPGAs were (and old MPGAs certainly were)
A typical fpga is a sea of logic units connected by some form of interconnect system. A logic unit typcially contains one or more flip-flops (used as latches and registers) and a lookup table system used to implement logic functions. They often also contain dedicated carry chain logic (addition and subtraction are pretty common in digital design and having dedicated logic for the carry chains can really speed them up).
The interconnect system will generally have a small number of dedicated clock distribution paths as well as the normal data paths to keep clock skew down (clock skew is bad).
Most higher end FPGAs have some dedicated stuff such as PLLs, multipliers etc.
How about learning how to program some small microcontroller (pic18 series are a nice choice, reasonablly clean architecture, nice package styles, tollerable development tools, free version of the C compiler availible with only minor crippling).
It is a whole different world and learning to deal with systems that are highly resourse limited is a refreshing change from the desktop mentality of "who cares about the bloat just wait for moores law to catch up".
right, the gp was discussing availibility not support though.
Larger buisnesses and those who have access to copies of thier buisnesses software (either legitimately under work at home provisions or otherwise) can continute to deploy windows XP for basically as long as they want. Volume licenses come with downgrade rights that run a long way back and the volume license edition of XP requires no activation.
For the home or small buisness user things look a lot less rosy. IIRC vista buisness OEM does come with downgrade rights to XP pro but depending on your situation it can be a pain to excercise them and if you bought a machine with a home edition of vista and you don't have access to volume licensing you will have to find old stock, pirate or bend the rules on whitebox OEM packs since retail and retail upgrade editions DO NOT come with downgrade rights.
If you aren't big enough to be multihoming yourself and you want reliable connectivity to as much of the internet as possible then your best bet is to try and find an ISP large enough to be multihomed but small enough not to be playing theese games and preferablly to listen to you as a customer. Yes this will cost a bit more, yes you get what you pay for.
Afaict cogents buisness model is selling hosting bandwidth dirt cheap. As such it makes sense they would stop all traffic to an ISP they want to peer with in order to try and push that ISP into peering with them through complaints from it's customers. Cogents aproach to trying to get better deals from other ISPs is similar to SPEWs approach to trying to stop ISPs being so friendly to spammers.
One of the restrictions is to allow a specific whitelist of software to run, specified as file hashes. Out of interest is this a secure hash algorithm like md5 or sha1 or is it something like crc32 which is trivial to create collisions for?
looks like ono is pure java so porting it to apps that aren't written in java (which lets be honest isn't the most popular choice for widely deployed desktop apps) means either a rewrite or a lot of work with the jni invocation apis and dragging all the bloat of the jvm into your app.
mmm, I have a macbook (which spends most of it's time running linux)and in may ways it is a nice machine but the combination of no hard drive light and a very quiet hard drive (I have a 200gb 4800rpm drive in there) means that sometimes I end up putting my ear to the machine just to try and tell whether there is disk activity or not.
I won't buy an iPhone as long as they're tethered and unsubsidized (full price WITH a contract?!), Umm AT&T is "subsidising" the iPhone, the mechanism is slightly different from with most phones but it most certainly is there.
Have you seen how much a legitimate unlocked iphone costs? (you can't buy them officially in the US but you can in some other countries)
I am using a machine here with intel integrated graphics (so no propietry drivers), loading google earth will without fail cause X to crash on this machine which is annoying to say the least (other 3D apps work albiet slowly).
wine also seems to have a nasty habbit of bringing down X from time to time though I can't reproduce that on demand:(.
The X server seems like it could really do with some improvement in it's resistance to poorly coded applications.
I have done a few clean windows installs in my time. Always from standard microsoft system builder (white box OEM) or volume license media. Versions from windows 95 through to vista.
As long as you know what hardware is in the box and have the original driver CDs to hand a clean windows install is pretty easy. Time consuming because of all of the reboots and disc swapping but pretty easy. If you don't then things can get a lot more painfull. Linux tends to find more hardware from the off but the stuff that doesn't work is often much harder to get working (wireless is a paritcular sore point).
The vista installs I did went very smoothly even though I didn't have any manufacturers driver disks for the machine but that is expected as it was a newer OS on older hardware so all the drivers are integrated.
The XP install on my macbook was one of the nicest windows installs i've ever done but I attribute that more to apple than to microsoft.
Lukilly I have avoided disk controller related problems so far (my machines have been too low end for scsi and other than the macbook and a sata card I added to another machine for a non-boot drive too old for sata) but from what I can gather theese are a major issue for installing both windows and linux. My understanding is that the versions of linux linux distros people want to use are usually more recent than the versions of windows people want to use so less likely to have problems. On the other hand windows is much easier to deal with if you do need to manually add a driver (the debian installer for example supports custom driver floppies but it doesn't seem to be documented anywhere how to build them).
I just wish they would do a better job of making sequals after the original developer quits.
Take spyro for example, not really my sort of game but my brother loved the three titles by the original developers.
Then insomniac games moved on (to create ratchet and clank, another great series at least for the first three games I haven't played gladiator, size matters and tolls of destruction yet) and whoevever owned the rights to the series got some other developers to create "enter the dragonfly". This games features total control change (and they didn't have the excuse of a different control pad, the dualshock and dualshock 2 are virtually identical), major storyline inconsistancies and horrible load times.
The problem I had with sim city 4 was that at least in my experiance it slowed down like hell once your city got over a certain size with things like the traffic analysis taking ages to update.
OTOH my newest gaming capable (read: has a dedicated graphics card) machine is just over 3 years old now (though the graphics card is newer, the machine wasn't initially bought for gaming) and was pretty low spec at the time so this may not be such an issue on modern hardware.
The problem as I understand it is that "BBC trust" (who are some kind of regulatory body with power over the BBC) have mandated "DRM". Obviously someone at the BBC thought that scanning for the user agent of something that couldn't save the video was adequate "DRM".
It is all rather stupid given that the BBC broadcast all thier content over unencrypted DVB-T and DVB-S anyway.
For as much Windows and Microsoft bashing that goes on in this community, it sure is funny to see how eager people are to get their hands on their latest beta. People do it for a variety of reasons, some for the ability to brag about it. Some because they want as much advanced warning of what MS has in store for us as they can. A few probablly want it because they actually want to use it but I strongly suspect they are in the minority.
you could but with many protocols you can't use the same account from multiple clients at once and even with protocols where you can getting duplicates of every message would get annoying rather quickly.
the traditional way to do a windowed gui was to limit each apps drawing area so that it could only draw within it's own window and force the app to redraw stuff when it's visibility changed. This system has the advantage of being light on ram and being low on CPU when windows aren't moving. However moving windows is a relatively expensive process both because of the need to ask apps to redraw and the need to actually move data arround in the screen buffer (this is why many older systems use a dotted box drawn with XOR to indicate window moving and only move the window when the user has chosen the final location). Also it is virtually impossible to support any kind of partial transparency or rotozooming under this system and even non rectangular windows are a pain.
3D games work in a totally different way. They work with a (large) set of textures and the scene is redrawn every frame building up from the back to the front and rotozooming everything into place. This makes transparency, drop shadows etc fairly easy and of course rotozooming is a fundamental requirement of a 3D game.
A compositing window system (afaict under X this requires support from both X itself and the window manager) draws each window into it's own buffer and then treats that as a texture. Then a frame for the screen is built up in much the same way a 3D game builds up a frame. This enables all sorts of effects from simple stuff like drop shadows and inverted colors to advanced stuff like a window selector that shows a thumbnail of each window or a desktop selector that puts the different desktops on the surface of a cube. Different window managers will obviously choose to use theese capabilities to different extents and in different ways.
Slashdot Mirror
Well debian already packaged the latest glibc in sid using gcc-4.3. That is how this issue was discovered in the first place.
do you know of anything that was broken besides SBCL (which was how this was discovered in the first place).
BTW I belive the intention of debian is to attack this problem from all sides. Afaict SBCL is being changed to keep the direction flag set for as short a time as possible. gcc is being changed to return to the older less likely to fail behaviour and linux is being changed to do what it should have done in the first place.
Well afaict the debian developers plan to modify gcc 4.3 so it behaves in the old way to reduce the risk of crashes when upgrading from one version of debian to the next. Dunno if gcc upstream will agree on that reasoning though. This isn't perfect though, even before gcc's behaviour changed there was still a risk that a signal handler would break the code that it interrupted.
Afaict this bug only affects a relatively small number of apps because little code messes with the direction flag in the first place
VHDL and verilog are a bit weired in the sense that they were originally designed as a method of describing hardware so it could be simulated and were later (ab)used for synthisis. This means that there are a lot of constructs that are perfectly legal verilog but will either cause an error in the synthisis tool or will synthisize to give different behaviour from thier simulation beaviour.
An FPGA is more like a vast sea (or field) of disconnected NAND NOR gates.
maybe some old FPGAs were (and old MPGAs certainly were)
A typical fpga is a sea of logic units connected by some form of interconnect system. A logic unit typcially contains one or more flip-flops (used as latches and registers) and a lookup table system used to implement logic functions. They often also contain dedicated carry chain logic (addition and subtraction are pretty common in digital design and having dedicated logic for the carry chains can really speed them up).
The interconnect system will generally have a small number of dedicated clock distribution paths as well as the normal data paths to keep clock skew down (clock skew is bad).
Most higher end FPGAs have some dedicated stuff such as PLLs, multipliers etc.
How about learning how to program some small microcontroller (pic18 series are a nice choice, reasonablly clean architecture, nice package styles, tollerable development tools, free version of the C compiler availible with only minor crippling).
It is a whole different world and learning to deal with systems that are highly resourse limited is a refreshing change from the desktop mentality of "who cares about the bloat just wait for moores law to catch up".
I presume he means using it on his own desktop at work so he is intimately familiar with it by the time the lusers get it and start demanding help.
right, the gp was discussing availibility not support though.
Larger buisnesses and those who have access to copies of thier buisnesses software (either legitimately under work at home provisions or otherwise) can continute to deploy windows XP for basically as long as they want. Volume licenses come with downgrade rights that run a long way back and the volume license edition of XP requires no activation.
For the home or small buisness user things look a lot less rosy. IIRC vista buisness OEM does come with downgrade rights to XP pro but depending on your situation it can be a pain to excercise them and if you bought a machine with a home edition of vista and you don't have access to volume licensing you will have to find old stock, pirate or bend the rules on whitebox OEM packs since retail and retail upgrade editions DO NOT come with downgrade rights.
If you aren't big enough to be multihoming yourself and you want reliable connectivity to as much of the internet as possible then your best bet is to try and find an ISP large enough to be multihomed but small enough not to be playing theese games and preferablly to listen to you as a customer. Yes this will cost a bit more, yes you get what you pay for.
Afaict cogents buisness model is selling hosting bandwidth dirt cheap. As such it makes sense they would stop all traffic to an ISP they want to peer with in order to try and push that ISP into peering with them through complaints from it's customers. Cogents aproach to trying to get better deals from other ISPs is similar to SPEWs approach to trying to stop ISPs being so friendly to spammers.
I was under the impression that nvidia just made the chipsets and a variety of vendors made the boards with varying heatsink designs.
One of the restrictions is to allow a specific whitelist of software to run, specified as file hashes.
Out of interest is this a secure hash algorithm like md5 or sha1 or is it something like crc32 which is trivial to create collisions for?
looks like ono is pure java so porting it to apps that aren't written in java (which lets be honest isn't the most popular choice for widely deployed desktop apps) means either a rewrite or a lot of work with the jni invocation apis and dragging all the bloat of the jvm into your app.
if you are feeling adventurous you could always try the release candidate of XP sp3 ( http://www.microsoft.com/downloads/details.aspx?FamilyId=114F3599-12AF-42B2-AAB1-B969A62C68A7&displaylang=en )
I would guess they will put out XP SP3 quietly after all the hooha about vista SP1 has died down.
mmm, I have a macbook (which spends most of it's time running linux)and in may ways it is a nice machine but the combination of no hard drive light and a very quiet hard drive (I have a 200gb 4800rpm drive in there) means that sometimes I end up putting my ear to the machine just to try and tell whether there is disk activity or not.
I won't buy an iPhone as long as they're tethered and unsubsidized (full price WITH a contract?!),
Umm AT&T is "subsidising" the iPhone, the mechanism is slightly different from with most phones but it most certainly is there.
Have you seen how much a legitimate unlocked iphone costs? (you can't buy them officially in the US but you can in some other countries)
XP professional x64 edition was never sold retail for some reason, you can still get system builder packs though.
first of all who uses foxytunes, I don't think i've ever see a firefox install with it in use.
secondly if MS wanted to cripple firefox they could do so far more easilly through say windows update.
I am using a machine here with intel integrated graphics (so no propietry drivers), loading google earth will without fail cause X to crash on this machine which is annoying to say the least (other 3D apps work albiet slowly).
:(.
wine also seems to have a nasty habbit of bringing down X from time to time though I can't reproduce that on demand
The X server seems like it could really do with some improvement in it's resistance to poorly coded applications.
I have done a few clean windows installs in my time. Always from standard microsoft system builder (white box OEM) or volume license media. Versions from windows 95 through to vista.
As long as you know what hardware is in the box and have the original driver CDs to hand a clean windows install is pretty easy. Time consuming because of all of the reboots and disc swapping but pretty easy. If you don't then things can get a lot more painfull. Linux tends to find more hardware from the off but the stuff that doesn't work is often much harder to get working (wireless is a paritcular sore point).
The vista installs I did went very smoothly even though I didn't have any manufacturers driver disks for the machine but that is expected as it was a newer OS on older hardware so all the drivers are integrated.
The XP install on my macbook was one of the nicest windows installs i've ever done but I attribute that more to apple than to microsoft.
Lukilly I have avoided disk controller related problems so far (my machines have been too low end for scsi and other than the macbook and a sata card I added to another machine for a non-boot drive too old for sata) but from what I can gather theese are a major issue for installing both windows and linux. My understanding is that the versions of linux linux distros people want to use are usually more recent than the versions of windows people want to use so less likely to have problems. On the other hand windows is much easier to deal with if you do need to manually add a driver (the debian installer for example supports custom driver floppies but it doesn't seem to be documented anywhere how to build them).
I just wish they would do a better job of making sequals after the original developer quits.
Take spyro for example, not really my sort of game but my brother loved the three titles by the original developers.
Then insomniac games moved on (to create ratchet and clank, another great series at least for the first three games I haven't played gladiator, size matters and tolls of destruction yet) and whoevever owned the rights to the series got some other developers to create "enter the dragonfly". This games features total control change (and they didn't have the excuse of a different control pad, the dualshock and dualshock 2 are virtually identical), major storyline inconsistancies and horrible load times.
The problem I had with sim city 4 was that at least in my experiance it slowed down like hell once your city got over a certain size with things like the traffic analysis taking ages to update.
OTOH my newest gaming capable (read: has a dedicated graphics card) machine is just over 3 years old now (though the graphics card is newer, the machine wasn't initially bought for gaming) and was pretty low spec at the time so this may not be such an issue on modern hardware.
The problem as I understand it is that "BBC trust" (who are some kind of regulatory body with power over the BBC) have mandated "DRM". Obviously someone at the BBC thought that scanning for the user agent of something that couldn't save the video was adequate "DRM".
It is all rather stupid given that the BBC broadcast all thier content over unencrypted DVB-T and DVB-S anyway.
For as much Windows and Microsoft bashing that goes on in this community, it sure is funny to see how eager people are to get their hands on their latest beta.
People do it for a variety of reasons, some for the ability to brag about it. Some because they want as much advanced warning of what MS has in store for us as they can. A few probablly want it because they actually want to use it but I strongly suspect they are in the minority.
you could but with many protocols you can't use the same account from multiple clients at once and even with protocols where you can getting duplicates of every message would get annoying rather quickly.
the traditional way to do a windowed gui was to limit each apps drawing area so that it could only draw within it's own window and force the app to redraw stuff when it's visibility changed. This system has the advantage of being light on ram and being low on CPU when windows aren't moving. However moving windows is a relatively expensive process both because of the need to ask apps to redraw and the need to actually move data arround in the screen buffer (this is why many older systems use a dotted box drawn with XOR to indicate window moving and only move the window when the user has chosen the final location). Also it is virtually impossible to support any kind of partial transparency or rotozooming under this system and even non rectangular windows are a pain.
3D games work in a totally different way. They work with a (large) set of textures and the scene is redrawn every frame building up from the back to the front and rotozooming everything into place. This makes transparency, drop shadows etc fairly easy and of course rotozooming is a fundamental requirement of a 3D game.
A compositing window system (afaict under X this requires support from both X itself and the window manager) draws each window into it's own buffer and then treats that as a texture. Then a frame for the screen is built up in much the same way a 3D game builds up a frame. This enables all sorts of effects from simple stuff like drop shadows and inverted colors to advanced stuff like a window selector that shows a thumbnail of each window or a desktop selector that puts the different desktops on the surface of a cube. Different window managers will obviously choose to use theese capabilities to different extents and in different ways.