Jetdirect are not very good under load. If you have multiple computers trying to print to the same JD box, it will only accept one connection at a time. and it will occasionally crash. It's been two years since I last admin'ed one of the JD, so maybe the new generation is better, but for rock-solid performance, I can only recommend AXIS. Easy to set up from almost any OS, can keep all its ports busy at all times.
You could do it with the ISDN4Linux programs and some scripting...4 or 5 ISDN PCI cards would give you 10 lines. I4L has DTMF support built in and an answering machine that actually works very well.
Nobody forces you to give anything away. Release your software under another license, and you don't have to release the source code. It's your code after all. Seems perfectly clear to me.
I hope it handles better under load than their Jetdirect boxes: these suck big time. Single tasking, no administrative control except for some cheesy Windows app, crashes under load, no security, no flexibilty in setup.
Sorta like their printers lately: flimsy and easy to break.
This "makes working on shared projects (in a +s directory) slightly easier" thing is an absolute myth. In order to do what you describe, the user would both have to a) have an idea of how to run chmod and b) actually _run_ chmod, at which point they can change the safe, secure, hopefully default permissions that allow _only_ the original user access to it.
Why would they need chmod? That's the nice thing about the usergroup scheme: it works without additional intervention and with no relaxing of security.
To recap: User home directories are ug+rwx with guid = uid. (umask=007) So all the files created in that home directory are only accessible to that user. Project directories are set g+rwxs, and guid=projectname. Now everyone in that project has access to that directory and every file that is created in that dir is automatically given the right permissions.
And the scenario where someone hacks into/etc/groups, well if they can hack groups, what's stopping them from hacking/etc/passwd?
I don't expect the RIAA to give me a new CD when mine gets a skip.
Ah, but you should. After all, you paid for a license to play the recording. So if your CD is scratched, it should be legal to make a CDR copy, or else have the record shop replace it for a handling+materials fee.
If you buy Photoshop, your license to use Photoshop is not tied to the CD, and you can get a replacement in case of loss/damage.
A single band GSM will serve you well in Europe. Coupled with a ultraportable like a Psion 5MX, will let you surf on the beach.
But you don't want to do this for serious work: it is horribly slow (9600bps), the latency is even worse and the telephone bill is going to be... entertaining. I know.... I use it that way when I am on holiday.
Using landlines, you need an ISP with an international presence. IBM Global Network is the only one I think with a wide enough coverage and no nonsense roaming on the standard account. ( maybe a small surcharge)
And bad battery life: it works an hour and then it drops dead. It won't save images without compression. No CompactFlash (Memorystick is a SONY proprietary format == more expensive and harder to find than CF,and of course a nice Sony lock-in effect) Sony isn't well known for releasing specs either. So under Linux, you are limited to serial transfer.
But jesus, are we still in the days of VAX? Big servers are great and it's useful for people to share them, but everyone ALSO has a decent desktop box.Why on earth do we need our computer to talk to itself through networking calls just to render a window?
Rest assured, it doesn't use networking calls if it is not necessary. It uses IPC (unix domain sockets,shared memory) where possible.
Does anyone have any figures for how much slower the X server system is than a direct windowing API? If it's a signifigant speed issue (and I imagine it is), we're going to have to do something about it.
And how are you going to measure that? How are you going to factor out driver optimisations? BTW, what you call a direct windowing API means moving the graphics subsystem into the kernel. It would save context switches. Any other design where a separate process handles the windowing and the display, has been done: it's called X.
How difficult would it be to abstract communication layer of the X server so that if it were running locally, it used direct calls, and if remotely, then by networking like it does now? Best of Both Worlds.
Depends on what you mean with "direct calls". A "direct call" into a shared library to draw a window can be handled in two ways:
A. either your graphics subsystem is in the kernel, and the code in the shared library thunks into the kernel for the window drawing code.
B. Your graphics subsystem is in a separate process, and the shared library marshalls the arguments of the function call into a structure and sends that structure via IPC to the display process. (Contrary to the belief of some, you can't just call "direct" into another process)
So there you have it: A = WinXX and B=XWindow. Take your pick. I guess moving the display subsystem into the kernel won't be very popular.
It's very simple to add it later, as the X servers for other OSes prove, and much more efficient.
X servers for other OS's don't prove anything (after all they are using the built-in network transparency of X). Let's take a real world example: remote viewing of NT sessions. The NT GDI was not written with network transparency in mind. So if it is, as you claim, " very simple" to add it later, why are all the solutions so clunky? VNC, PC Anywhere and Citrix don't even come close to the smoothness of X.
Why do X applications need to convert all GUI-API calls to a protocol suitable for network transport, send the data over a Unix domain socket and convert it back to internal API calls for the X server?
Why not? The overhead is small. IPC is pretty efficient. The only way to get more efficient is to move the display system into the kernel (ala Winxx and WinNT). It would save you context switches. If you run your display driver/Windowing system in a separate process, you still have to pass your command via IPC ( domain socket, shared memory).
I must have one magical motherboard then. Using the SMP stubs in the Linux kernels on my dual-PII, I get twice the rc5des output than a single processor. Funny, isn't it?
>>but why is it that when talking about user interfaces, Windows kicks Linux's ass.
Linux has a user interface tailored to its users: clean, powerful and very expressive. So I think a lot of people would disagree with your statement.
And as we are talking about VB here: you can smell a VB app from miles away: clunky , arbitrary limitations and UI elements that are used in totally inapropriate ways.
Look, UI design is hard.... no shortcuts. It takes experience , an open mind and lots of hard work.
>> It Linux *were* clearly superior to Windows then the techies would be switching over just for the sheer joy of it.
But they are.... Most of my friends in development are longing for the day they can ditch Windows. Reason: ease of administration, flexibility and stability. Alas, there is the small matter of economics: basically they develop for the platform that the potential customers have.
Using a time machine is obvious. Developing one isn't. As nobody even has a clue how to develop one, I'd rate a time machine very patentable.
If something seems obvious once you read it, it is because it was obvious, but nobody encountered a problem which demanded that solution before. Should that be patentable? Well, the objective of the patent system is to grant a monopoly in return for full disclosure. What is the disclosure worth to society if all it takes to duplicate the "invention", is someone encountering the same problem and coming up with the same solution?
So the benchmark for a patent should be: if someone encounters the problem and comes up with the same solution without prior knowledge of the patent in question, it is unpatentable. Something like clean-room reverse-engineering. This rules out this graffiti patent ( pretty obvious ), all gene sequencing ( if you don't know what it does, and you didn't even create it yourself, why should you have a patent?)
Now some people will object that there is a cost associated with DNA sequencing and that someone who maps a gene should be able to recoup their costs via a patent. There are a lot of very similar activities in this category: a census, drawing maps of an area, description and classification of flora. None of them are patentable, and yet somebody out there is doing it. (There is an open-source theme here....)
But back on topic: modifying an alphabet so machine recognition is enhanced is an application of a common pattern in software development: OCR fonts and barcodes come to mind. So should the technique as it applies to human handwriting be patentable? Of course not. In the context of computer science, it is an obvious "invention".
BTW, Alphabets mean what you want them to mean: you mold them until they fit: if all you have is a rock and a chisel, you'll find that runes and roman numerals are easier than our current alphabet. If all you have is a limited resolution touchpad and a slow processor, you modify the alphabet to maximise the distance between each character.
HP Jetdirect print servers are very bad. They don't hold up under load, can crash when two computers send a job at the same time, no tools for management.
Get an Axis: robust engineering, very good print management.
Most of the population in France can't speak anything but french. It's a bit better than Spain, a lot better than Italy, on a par with Germany. Nothing like the Netherlands (English), Belgium (Flanders: french+english) or the scandinavian countries (English).
There are so many ill-informed claims and errors in this post , I don't know where to begin.
1. the fastest tapes are not around 1MB/s. In the low-medium price range, there is DDS-4, AIT, and other tape technologies: Size: upto 40Gb, 3MB/s (sustained,uncompressed) Let's spec out a DDS-4 solution: 20Gb,3MB/s native. Street price $1300. Hook up 20 of those in a library and you have 400Gb capacity with 60Mb/s throughput.
2. Same price range as the RAID array, there are tape drives (AMPEX) that go to 15Mb/s (sustained,uncompressed) with sizes upto 330Gb.
3. You are getting 50MB/s from an EIDE drive? Last time I looked, a high-end SCSI drive had only 300+ Mbit/s internal transfer rate with sustained transfers of round 20MB/s. And you have an EIDE drive that goes up to 400Mbit/s internal?????
Network access is a right if you pay for it. Those students paid for it.
It's like your ISP deciding to view what kind of servers their customers have running and what files are accessible.
Or your landlord insisting that the only internet access your apartment is going to get goes through his Cybernanny-firewall at $100/month. No telephone lines, no DSL or cable modems allowed.
Slashdot Mirror
They don't mention that the telnet interface is by default only accessible from the inside of the network.
Jetdirect are not very good under load. If you have multiple computers trying to print to the same JD box, it will only accept one connection at a time. and it will occasionally crash. It's been two years since I last admin'ed one of the JD, so maybe the new generation is better, but for rock-solid performance, I can only recommend AXIS. Easy to set up from almost any OS, can keep all its ports busy at all times.
Edit -> Preferences -> Navigator -> Tabbed browsing will let you use middle-click for tabbed browsing
the 440BX doesn't have a MTH. What you are talking about is probably the Camino chipset,the 820.
You could do it with the ISDN4Linux programs and some scripting...4 or 5 ISDN PCI cards would give you 10 lines. I4L has DTMF support built in and an answering machine that actually works very well.
Nobody forces you to give anything away. Release your software under another license, and you don't have to release the source code. It's your code after all. Seems perfectly clear to me.
I hope it handles better under load than their Jetdirect boxes: these suck big time. Single tasking, no administrative control except for some cheesy Windows app, crashes under load, no security, no flexibilty in setup.
Sorta like their printers lately: flimsy and easy to break.
To recap: User home directories are ug+rwx with guid = uid. (umask=007) So all the files created in that home directory are only accessible to that user. Project directories are set g+rwxs, and guid=projectname. Now everyone in that project has access to that directory and every file that is created in that dir is automatically given the right permissions.
And the scenario where someone hacks into /etc/groups, well if they can hack groups, what's stopping them from hacking /etc/passwd?
If you buy Photoshop, your license to use Photoshop is not tied to the CD, and you can get a replacement in case of loss/damage.
NT is not a microkernel OS. The OS/2, DOS and POSIX "servers" are just subsystems, something like Winelib on Linux.
How "micro" is a "microkernel" that includes a graphics subsystem?
A single band GSM will serve you well in Europe. Coupled with a ultraportable like a Psion 5MX, will let you surf on the beach.
... entertaining. I know.... I use it that way when I am on holiday.
But you don't want to do this for serious work: it is horribly slow (9600bps), the latency is even worse and the telephone bill is going to be
Using landlines, you need an ISP with an international presence. IBM Global Network is the only one I think with a wide enough coverage and no nonsense roaming on the standard account. ( maybe a small surcharge)
It won't save images without compression.
No CompactFlash (Memorystick is a SONY proprietary format == more expensive and harder to find than CF
Sony isn't well known for releasing specs either. So under Linux, you are limited to serial transfer.
Here is a review of the Sony vs an Olympus.
So there you have it: A = WinXX and B=XWindow. Take your pick. I guess moving the display subsystem into the kernel won't be very popular.
I must have one magical motherboard then. Using the SMP stubs in the Linux kernels on my dual-PII, I get twice the rc5des output than a single processor. Funny, isn't it?
>>but why is it that when talking about user interfaces, Windows kicks Linux's ass.
Linux has a user interface tailored to its users: clean, powerful and very expressive. So I think a lot of people would disagree with your statement.
And as we are talking about VB here: you can smell a VB app from miles away: clunky , arbitrary limitations and UI elements that are used in totally inapropriate ways.
Look, UI design is hard.... no shortcuts. It takes experience , an open mind and lots of hard work.
>> It Linux *were* clearly superior to Windows then the techies would be switching over just for the sheer joy of it.
But they are.... Most of my friends in development are longing for the day they can ditch Windows. Reason: ease of administration, flexibility and stability. Alas, there is the small matter of economics: basically they develop for the platform that the potential customers have.
>>Ever have windows crash in safe mode? I haven't.
I have. Repeatedly.
Using a time machine is obvious. Developing one isn't. As nobody even has a clue how to develop one, I'd rate a time machine very patentable.
If something seems obvious once you read it, it is because it was obvious, but nobody encountered a problem which demanded that solution before. Should that be patentable? Well, the objective of the patent system is to grant a monopoly in return for full disclosure. What is the disclosure worth to society if all it takes to duplicate the "invention", is someone encountering the same problem and coming up with the same solution?
So the benchmark for a patent should be: if someone encounters the problem and comes up with the same solution without prior knowledge of the patent in question, it is unpatentable. Something like clean-room reverse-engineering. This rules out this graffiti patent ( pretty obvious ), all gene sequencing ( if you don't know what it does, and you didn't even create it yourself, why should you have a patent?)
Now some people will object that there is a cost associated with DNA sequencing and that someone who maps a gene should be able to recoup their costs via a patent. There are a lot of very similar activities in this category: a census, drawing maps of an area, description and classification of flora. None of them are patentable, and yet somebody out there is doing it. (There is an open-source theme here....)
But back on topic: modifying an alphabet so machine recognition is enhanced is an application of a common pattern in software development: OCR fonts and barcodes come to mind. So should the technique as it applies to human handwriting be patentable? Of course not. In the context of computer science, it is an obvious "invention".
BTW, Alphabets mean what you want them to mean: you mold them until they fit: if all you have is a rock and a chisel, you'll find that runes and roman numerals are easier than our current alphabet. If all you have is a limited resolution touchpad and a slow processor, you modify the alphabet to maximise the distance between each character.
HP Jetdirect print servers are very bad. They don't hold up under load, can crash when two computers send a job at the same time, no tools for management.
Get an Axis: robust engineering, very good print management.
Most of the population in France can't speak anything but french. It's a bit better than Spain, a lot better than Italy, on a par with Germany. Nothing like the Netherlands (English), Belgium (Flanders: french+english) or the scandinavian countries (English).
BTW, Germany is east of France.
There are so many ill-informed claims and errors in this post , I don't know where to begin.
1. the fastest tapes are not around 1MB/s. In the low-medium price range, there is DDS-4, AIT, and other tape technologies: Size: upto 40Gb, 3MB/s (sustained,uncompressed) Let's spec out a DDS-4 solution: 20Gb,3MB/s native. Street price $1300. Hook up 20 of those in a library and you have 400Gb capacity with 60Mb/s throughput.
2. Same price range as the RAID array, there are tape drives (AMPEX) that go to 15Mb/s (sustained,uncompressed) with sizes upto 330Gb.
3. You are getting 50MB/s from an EIDE drive? Last time I looked, a high-end SCSI drive had only 300+ Mbit/s internal transfer rate with sustained transfers of round 20MB/s. And you have an EIDE drive that goes up to 400Mbit/s internal?????
Network access is a right if you pay for it. Those students paid for it.
It's like your ISP deciding to view what kind of servers their customers have running and what files are accessible.
Or your landlord insisting that the only internet access your apartment is going to get goes through his Cybernanny-firewall at $100/month. No telephone lines, no DSL or cable modems allowed.
Apparently, the static pages are served by NT and the perl scripts by Apache/Solaris.