I have found usb to serial adapters that were unable to send BREAK signals correctly (needed to escape to the prom on servers, and to access recovery modes on some cisco devices)... As terminal emulator i was using Minicom, which most certainly can send break signals when using a hardware serial port. I've also encountered some devices which work well on Linux, but have various broken functionality (like break signals) on windows or mac (typically the linux drivers are open source and get fixed, while windows/mac has crufty closed source drivers created by the manufacturer)..
Another example, is the FTDI drivers... If you disconnect an FTDI serial device from a mac while you have the port in use, using their official drivers, you get a kernel panic. Linux handles this situation fine.
I think his point was that you still need code to drive your ethernet card, and you need to implement some kind of protocol over it even if not tcp/ip.. And if you use a nonstandard protocol (ie not telnet over tcp) then you will also need to implement a client for it.
And some vendors still release their firmware updates as floppy images... Even worse when they come as a windows program, which when executed writes the image to a floppy (ie you don't just have the raw image file which you could dd on linux or load into the virtual floppy feature in most lights out management firmware).
Serial is standard, whereas USB devices typically need drivers... The Marvell Sheevaplug for instance has a USB console port, but you need to install drivers... Linux has these drivers by default, but osx/windows don't, and the osx drivers are horrendously buggy.
Hiding your SSID can actually be detrimental... If your SSID is open, then your machine can see its broadcasts and connect to it... If the SSID is hidden, then your machine has to probe for it by name.. Meaning that if your machine is away from its usual location, you can see what network its looking for...
If the SSID is hidden, then someone trying to break into it just needs to sniff traffic for a while to get the SSID anyway.
They don't care about the real pirates, there's nothing they can do about the hardcore crackers anyway, and these people would rather do without than pay for software.
What they do care about is the grey market middle ground, the kid who buys a copy from a store but lets his friends borrow it, the people who resell used games... These people, unlike the hardcore pirates, *ARE* willing to spend money on games, so the games companies seek to extract as much of it out of them as possible via whatever means necessary.
You can also modify the binaries to ignore a man in the middle, and establish a connection with your proxy quite easily... Encryption is pretty easily broken when you hold the keys.
What they're more likely to do, is make a caching proxy server so as people play they will gradually build up all of the necessary content and ubisoft will just see 1 connection and assume it to be whoever legitimately bought the game prior to making the initial warez copy...
My DSL used to be stable like yours too, up until the last few months when i believe the telco "upgraded" their dslams... Since then, i typically get disconnections (loss of sync, forcing router to retrain) every few hours especially in the afternoons and evenings when i can sometimes get 3-4 disconnections in an hour, and my average sync rate is 1/3 of what it used to be. And there are simply no other options here aside from dialup, which is billed by the minute and obviously very slow.
Legit purchases have been known to come with malware too, there have been various cases of storage devices being shipped out with malware preinstalled for instance.
I imagine most warez groups will be quite insulted to have their work branded as counterfeiting... Counterfeit goods are typically cheap (often inferior) copies which are falsely sold as originals... Warez on the other hand, at least the kind you download, is quite clearly labelled as warez and often branded by the group who ripped it, and is usually a superior product to the original work as the warez copies have drm schemes and other nasties removed.
Problems getting the wrong mouse button is most likely the reason Apple have traditionally only used a single button...
And a voice recognition system for interacting the computer would mean a return to the commandline, a commandline is effectively a conversation in textual form between you and the computer, where the computer has it's own simplified language. It would be extremely difficult and unnatural to try and explain and interact with gui elements vocally.
You get broadband, subsidised laptop and support... You're not just paying for the laptop, you're actually paying for a managed service... That's actually a good thing, many people simply lack the required knowledge to manage a full featured OS connected to the internet.
People think they need antivirus, and that it is an essential part of running a computer... Give such people a mac or linux box and they will ask for antivirus for it...
The windows 9x legacy goes much further than that... A lot of the security problems in windows as a result of code that was ported over from 9x or needed to be compatible with the 9x way of doing things... The NT kernel security model was pretty good, but a lot of the cruft microsoft brought over from the 9x series renders much of it useless.
And macos has similar problems... I have a TV here which is 1080p capable, but it incorrectly reports its capabilities... I can't get windows or macos to do more than 1080i, but linux can be forced to 1080p quite easily.
I don't see why there's even any argument, it's obvious that text files are a vastly superior method of storing configuration...
They're human readable, easily searchable, trivially editable from recovery environments (ie dont require any specialised tools), trivially editable over a serial console, trivially parsed or generated with simple scripts, can be easily backed up in revision control systems and diffed etc, you can easily put examples online and textfiles will usually allow you to put comments in the files which can be extremely useful.
Are there any possible advantages to having proprietary binary config files?
A lot (pretty much most) of unix programs, especially open source ones, will generally let you configure where they install at compiletime... It's typically only closed source stuff that makes strange decisions about where it wants to install.
Backups on windows (and earlier linux versions, not tried the latest ubuntu etc) are a hassle for end users, who end up either manually copying files (and often forget to do so)... My grandparents have a mac, and i got them an external drive and have time machine configured for automatic backups.
This is why many people don't like closed source proprietary software... The original vendor of this software has stopped developing or distributing it, this would be bad enough and effectively turn existing versions into abandonware... But given Apple's distribution model, this software is now effectively completely defunct. What happens to all the people who paid for the non free version?
The article mentions that they use a program of their own creation for monitoring memory usage, so how the users interpret the data is irrelevant as the program will send what it believes to be correct. Wether the program is accurate or not is another matter, but the fact it doesn't report every system as using 100% of its memory suggests it is at least somewhat aware of superfetch etc.
The fonts should be perfectly readable so long as your screen informs X of its physical dimensions, so X can work out the appropriate DPI setting... Infact, the text should then be the same physical size regardless of resolution or screen size. The problem is that some laptop panels don't do this, so X cannot work out the DPI and uses a default, a default which is usable but not ideal for typical size CRT screens but unsuitable for very small netbook screens.
Slashdot Mirror
I have found usb to serial adapters that were unable to send BREAK signals correctly (needed to escape to the prom on servers, and to access recovery modes on some cisco devices)...
As terminal emulator i was using Minicom, which most certainly can send break signals when using a hardware serial port.
I've also encountered some devices which work well on Linux, but have various broken functionality (like break signals) on windows or mac (typically the linux drivers are open source and get fixed, while windows/mac has crufty closed source drivers created by the manufacturer)..
Another example, is the FTDI drivers... If you disconnect an FTDI serial device from a mac while you have the port in use, using their official drivers, you get a kernel panic. Linux handles this situation fine.
I think his point was that you still need code to drive your ethernet card, and you need to implement some kind of protocol over it even if not tcp/ip..
And if you use a nonstandard protocol (ie not telnet over tcp) then you will also need to implement a client for it.
I believe the write protect switches in SD cards are implemented in software, that is you can force the hardware to write to them anyway.
And some vendors still release their firmware updates as floppy images... Even worse when they come as a windows program, which when executed writes the image to a floppy (ie you don't just have the raw image file which you could dd on linux or load into the virtual floppy feature in most lights out management firmware).
Yes but are you sure what the ports are for?
They could be to let people load firmware updates from usb sticks for instance.
Serial is standard, whereas USB devices typically need drivers...
The Marvell Sheevaplug for instance has a USB console port, but you need to install drivers... Linux has these drivers by default, but osx/windows don't, and the osx drivers are horrendously buggy.
Hiding your SSID can actually be detrimental...
If your SSID is open, then your machine can see its broadcasts and connect to it... If the SSID is hidden, then your machine has to probe for it by name.. Meaning that if your machine is away from its usual location, you can see what network its looking for...
If the SSID is hidden, then someone trying to break into it just needs to sniff traffic for a while to get the SSID anyway.
They don't care about the real pirates, there's nothing they can do about the hardcore crackers anyway, and these people would rather do without than pay for software.
What they do care about is the grey market middle ground, the kid who buys a copy from a store but lets his friends borrow it, the people who resell used games...
These people, unlike the hardcore pirates, *ARE* willing to spend money on games, so the games companies seek to extract as much of it out of them as possible via whatever means necessary.
You can also modify the binaries to ignore a man in the middle, and establish a connection with your proxy quite easily...
Encryption is pretty easily broken when you hold the keys.
What they're more likely to do, is make a caching proxy server so as people play they will gradually build up all of the necessary content and ubisoft will just see 1 connection and assume it to be whoever legitimately bought the game prior to making the initial warez copy...
My DSL used to be stable like yours too, up until the last few months when i believe the telco "upgraded" their dslams...
Since then, i typically get disconnections (loss of sync, forcing router to retrain) every few hours especially in the afternoons and evenings when i can sometimes get 3-4 disconnections in an hour, and my average sync rate is 1/3 of what it used to be.
And there are simply no other options here aside from dialup, which is billed by the minute and obviously very slow.
Legit purchases have been known to come with malware too, there have been various cases of storage devices being shipped out with malware preinstalled for instance.
I imagine most warez groups will be quite insulted to have their work branded as counterfeiting... Counterfeit goods are typically cheap (often inferior) copies which are falsely sold as originals...
Warez on the other hand, at least the kind you download, is quite clearly labelled as warez and often branded by the group who ripped it, and is usually a superior product to the original work as the warez copies have drm schemes and other nasties removed.
Problems getting the wrong mouse button is most likely the reason Apple have traditionally only used a single button...
And a voice recognition system for interacting the computer would mean a return to the commandline, a commandline is effectively a conversation in textual form between you and the computer, where the computer has it's own simplified language. It would be extremely difficult and unnatural to try and explain and interact with gui elements vocally.
If you did implement voice recognition, you would have to return to a command line environment... Talking is much closer to a command line.
You get broadband, subsidised laptop and support... You're not just paying for the laptop, you're actually paying for a managed service... That's actually a good thing, many people simply lack the required knowledge to manage a full featured OS connected to the internet.
People think they need antivirus, and that it is an essential part of running a computer...
Give such people a mac or linux box and they will ask for antivirus for it...
The windows 9x legacy goes much further than that... A lot of the security problems in windows as a result of code that was ported over from 9x or needed to be compatible with the 9x way of doing things... The NT kernel security model was pretty good, but a lot of the cruft microsoft brought over from the 9x series renders much of it useless.
And macos has similar problems...
I have a TV here which is 1080p capable, but it incorrectly reports its capabilities... I can't get windows or macos to do more than 1080i, but linux can be forced to 1080p quite easily.
I don't see why there's even any argument, it's obvious that text files are a vastly superior method of storing configuration...
They're human readable, easily searchable, trivially editable from recovery environments (ie dont require any specialised tools), trivially editable over a serial console, trivially parsed or generated with simple scripts, can be easily backed up in revision control systems and diffed etc, you can easily put examples online and textfiles will usually allow you to put comments in the files which can be extremely useful.
Are there any possible advantages to having proprietary binary config files?
A lot (pretty much most) of unix programs, especially open source ones, will generally let you configure where they install at compiletime... It's typically only closed source stuff that makes strange decisions about where it wants to install.
Backups on windows (and earlier linux versions, not tried the latest ubuntu etc) are a hassle for end users, who end up either manually copying files (and often forget to do so)...
My grandparents have a mac, and i got them an external drive and have time machine configured for automatic backups.
This is why many people don't like closed source proprietary software...
The original vendor of this software has stopped developing or distributing it, this would be bad enough and effectively turn existing versions into abandonware... But given Apple's distribution model, this software is now effectively completely defunct. What happens to all the people who paid for the non free version?
The article mentions that they use a program of their own creation for monitoring memory usage, so how the users interpret the data is irrelevant as the program will send what it believes to be correct.
Wether the program is accurate or not is another matter, but the fact it doesn't report every system as using 100% of its memory suggests it is at least somewhat aware of superfetch etc.
The fonts should be perfectly readable so long as your screen informs X of its physical dimensions, so X can work out the appropriate DPI setting...
Infact, the text should then be the same physical size regardless of resolution or screen size. The problem is that some laptop panels don't do this, so X cannot work out the DPI and uses a default, a default which is usable but not ideal for typical size CRT screens but unsuitable for very small netbook screens.