You could look at Starnet's XWin32... I don't know whether it's any good, and it's also payware, but the latest versions support suspend/resume of sessions which seems to be what the poster wants.
The best way, in my opinion, for people to create an application that uses concurrency is to design it that way. I know that sounds trite, but it's true.
A simple example. If you start with a very large number of parallel processes, and wish to create a sequential version of them, the solution is so simple we delegate it to OS run-time in the form of the scheduler.
If you have a single sequential process, and wish to create a large number of parallel process, the problem is so difficult that, in the general case, you can't (although some compilers manage some parts of the job, and some processors manage some parts).
The formalism that has proved itself time and time again in getting parallel design right is Hoare's CSP, which promoted the idea of autonomous processes sending and receiving discrete messages to each other. The reasons for this include:
- a process' memory (state) cannot be changed without its explicit say-so (because messages must be accepted, not just sent).
- various properties ensure "WYSISYG", or compositional, programming - if you put two processes together that have been independedntly tested, you can be sure that their behaviour doesn't change just because you've put them together. This is not true of pthreads/winthreads (in general).
- because there is a formalism (CSP) behind implementations such as JCSP (http://en.wikipedia.org/wiki/JCSP) there are clear program transformational rules, which helps in many ways to make programs safer.
Do have a look...
One last point. Once you have a somewhat threaded[1] system, UI responsiveness is, on modern systems, mostly a function of program size. Large programs (including the OS) find it very difficult to be responsive because the CPU is being asked to access items all over memory. The reason that is bad is that a memory access that misses CPU cache incurs an enormous penalty - maybe as much as 1000 CPU cycles - during which the processor is often twiddling its thumbs. Reducing code bloat is essential to improve this, not increasing the number of threads.
[1] that is, tasks that take noticeable time are separated out.
I'm not supporting parallel movie players as a benchmark, but "most operating systems have efficient, low overhead threads" - Ha! All I can say is you have never seen "efficient threads". Not one of the major OSs have truly efficient threading. For that you have to look at the real time kernels, where it really counts, and to projects like jcsp (http://en.wikipedia.org/wiki/JCSP) where both thread startup and context switches take a few hundred cycles, not tens of thousands.
I would agree with the need to bear your audience in mind.
Writing good English is just as much a skill as writing good C# or Java. Indeed, you can't just say "good English", as the style matters too... good novel writing, tech docs, journalism, and academic writing are all different styles that must be learnt, and some people are better at one style than others.
Regarding tools, the one that keeps being mentioned as being the best for tech writing is FrameMaker. There is a caveat: it takes time and skill to set up properly. However, once that is done it can be used well by authors who aren't also geeks!
If you feel tempted by DocBook or one of the other XML based systems, I've found the XML Mind XML Editor decent, but with less flexibility than Frame.
MS Word should be ignored -- it is too unconstrained and too buggy for large documents (>100 pages).
Some people really like LaTeX, and it is indeed good even for large documents, but only if (a) your authors are prepared to learn it and (b) your desired output formats are supported. It is supported by Windows as well as Unix, and there are GUI front-ends (by default it is batch processed).
I believe that patents are not (any longer) necessary, or even good for, society.
Originally, a patent was granted on a single physical item, the invention and production of which was considered hard, even by those in the relevant field. Then, it was important that this was a finished work, not just an idea as to how the work could be completed.
The intention of this was to prevent the knowledge of how the invention worked from being secret and eventually dying with the inventor, by forcing disclosure in return for monopoly rights for a fairly limited period.
The idea of this system helping the small player came along later. It might even work if only the lawyers didn't get in the way: as it is now, virtually no small player can afford to challenge a patent infringement because they cannot fund the case costs (even if they win) because (a) there is no guarantee of winning and (b) from what I understand, case costs are not usually charged to the loser.
In my view, all of the benefits of patents to society are now lost: many patents are not actually for items that are of huge importance to future society -- viz the gif patent which, while useful and widely used, has at least 3 major competitors. And even for those that are truly useful, the patent period is now so long, and the pace of society so fast, that in many cases companies are forced to reinvent things anyway.
On the bad side, we have patent offices (in at least the USA and UK) that are not technically competent to answer the questions "is this inventive?", "is there propr art?", "have they actually created it?". Instead, they seem to be willing to pass almost anything that comes in the door, accepting things written in the broadest possible terms (intended to cover a multitude of items rather than the single item originally envisaged).
In summary, the patent system as currently implemented is fatally flawed, and I really believe that the world would be better without it. It is possible, although not probable, that some new system could or should be created replace it, but I have no faith in the value to society of patents, and therefore no reason for society (in the form of governments) should play a part in them.
Well, with the recent moves by Intel and AMD into multiprocessor chips, and the recent news about Pentuim-4 not being much faster than Pentium-III (Pentium-M) because of pipeline issues, I would have thought it clear that yes, the GHz war is on the wane.
Sure, we may have processors in the 4-5GHz range eventually but at the moment, I doubt that things will go a lot further, clock-speed wise. And even a 5HGz CPU is only about 30% faster than the current crop: hardly a big win!
The fact that it can be ported (which I dont disagree with) doesn't mean it would be good to port it for the average user.
Just think how many people now use XBox. It is basically a PC, and yet many gamers went to XBox because of all thr problems they had in getting their PC games to run on their particular hardware.
Platform stability is more important than some people believe.
Something about DVDs not lasting even 30 years, according to some people. OK, so you copy them to another DVD in 20 years, but how many will slip through the net and not be copied?
I have quite successfully used a Nikon LS30 scanner (8 bit 4000 dpi) on LInux using VueScan. It plugs into the Firewire port using the Kernel "SBP" firewire driver. VueScan does the rest.
Colour management is another issue. To get decent results using supplied (approximate) profiles is not good: use a profiling tool such as those sold by Gretag-Macbeth for about £120 ($250?)
But online bank accounts ?all? use https -- that is, using SSL as well. Surely that makes it a lot harder to get at useful information on the WEP link?
The thing that bothers me about predictions of "losses" such as these losses in excess of EUR4.5 billion annually due to piracy is that it assumes that the money was available to be paid. In many cases (e.g. children swapping games software) it simply isn't.
The reason is of course that multiplying number pirated * max selling price produced "shocking" numbers for the policy makers, even though it's almost certainly wildly inaccurate.
No. Security is designed into (each component of) a product, verified in the implementation, and then verified in the final system. Saying "MS SQL Server is secure" is complete nonsense as a stand-alone statement... secure against what? in what conditions?
R.
Re:Cable is Better in YOUR area
on
DSL Rising
·
· Score: 1
Last time I looked, Europe wasn't "a country".
And there are really quite a few broadcasters, including one of the best (good ol'e Beeb!).
I am not doing this, but it is something I would like to do; under the current 'pay for channel bundles' method, most channels piggy-back on the success of a few, while forcing me to 'buy' channels I have no interest in.
I must agree with those who view the posted questions as being a curious choice. Some of my own are:
1. Exactly what is gravity, and can it be controlled ?
2. Can/does anything travel faster than light, with or without wormholes?
3. How can power be generated in ecologically safe ways (that is, without unfortunate byproducts)?
4. How can people fly safely, quickly and affordably without requiring fossil fuels?
5. Is a true 3d hologram (a la Rimmer/Cdr Data/etc) possible?
Slashdot Mirror
You could look at Starnet's XWin32... I don't know whether it's any good, and it's also payware, but the latest versions support suspend/resume of sessions which seems to be what the poster wants.
The best way, in my opinion, for people to create an application that uses concurrency is to design it that way. I know that sounds trite, but it's true. A simple example. If you start with a very large number of parallel processes, and wish to create a sequential version of them, the solution is so simple we delegate it to OS run-time in the form of the scheduler. If you have a single sequential process, and wish to create a large number of parallel process, the problem is so difficult that, in the general case, you can't (although some compilers manage some parts of the job, and some processors manage some parts). The formalism that has proved itself time and time again in getting parallel design right is Hoare's CSP, which promoted the idea of autonomous processes sending and receiving discrete messages to each other. The reasons for this include: - a process' memory (state) cannot be changed without its explicit say-so (because messages must be accepted, not just sent). - various properties ensure "WYSISYG", or compositional, programming - if you put two processes together that have been independedntly tested, you can be sure that their behaviour doesn't change just because you've put them together. This is not true of pthreads/winthreads (in general). - because there is a formalism (CSP) behind implementations such as JCSP (http://en.wikipedia.org/wiki/JCSP) there are clear program transformational rules, which helps in many ways to make programs safer. Do have a look... One last point. Once you have a somewhat threaded[1] system, UI responsiveness is, on modern systems, mostly a function of program size. Large programs (including the OS) find it very difficult to be responsive because the CPU is being asked to access items all over memory. The reason that is bad is that a memory access that misses CPU cache incurs an enormous penalty - maybe as much as 1000 CPU cycles - during which the processor is often twiddling its thumbs. Reducing code bloat is essential to improve this, not increasing the number of threads. [1] that is, tasks that take noticeable time are separated out.
I'm not supporting parallel movie players as a benchmark, but "most operating systems have efficient, low overhead threads" - Ha! All I can say is you have never seen "efficient threads". Not one of the major OSs have truly efficient threading. For that you have to look at the real time kernels, where it really counts, and to projects like jcsp (http://en.wikipedia.org/wiki/JCSP) where both thread startup and context switches take a few hundred cycles, not tens of thousands.
I would agree with the need to bear your audience in mind.
Writing good English is just as much a skill as writing good C# or Java. Indeed, you can't just say "good English", as the style matters too... good novel writing, tech docs, journalism, and academic writing are all different styles that must be learnt, and some people are better at one style than others.
Regarding tools, the one that keeps being mentioned as being the best for tech writing is FrameMaker. There is a caveat: it takes time and skill to set up properly. However, once that is done it can be used well by authors who aren't also geeks!
If you feel tempted by DocBook or one of the other XML based systems, I've found the XML Mind XML Editor decent, but with less flexibility than Frame.
MS Word should be ignored -- it is too unconstrained and too buggy for large documents (>100 pages).
Some people really like LaTeX, and it is indeed good even for large documents, but only if (a) your authors are prepared to learn it and (b) your desired output formats are supported. It is supported by Windows as well as Unix, and there are GUI front-ends (by default it is batch processed).
HTH,
That link relates to RH Enterprise Linux, not Fedora Core.
I believe that patents are not (any longer) necessary, or even good for, society.
Originally, a patent was granted on a single physical item, the invention and production of which was considered hard, even by those in the relevant field. Then, it was important that this was a finished work, not just an idea as to how the work could be completed.
The intention of this was to prevent the knowledge of how the invention worked from being secret and eventually dying with the inventor, by forcing disclosure in return for monopoly rights for a fairly limited period.
The idea of this system helping the small player came along later. It might even work if only the lawyers didn't get in the way: as it is now, virtually no small player can afford to challenge a patent infringement because they cannot fund the case costs (even if they win) because (a) there is no guarantee of winning and (b) from what I understand, case costs are not usually charged to the loser.
In my view, all of the benefits of patents to society are now lost: many patents are not actually for items that are of huge importance to future society -- viz the gif patent which, while useful and widely used, has at least 3 major competitors. And even for those that are truly useful, the patent period is now so long, and the pace of society so fast, that in many cases companies are forced to reinvent things anyway.
On the bad side, we have patent offices (in at least the USA and UK) that are not technically competent to answer the questions "is this inventive?", "is there propr art?", "have they actually created it?". Instead, they seem to be willing to pass almost anything that comes in the door, accepting things written in the broadest possible terms (intended to cover a multitude of items rather than the single item originally envisaged).
In summary, the patent system as currently implemented is fatally flawed, and I really believe that the world would be better without it. It is possible, although not probable, that some new system could or should be created replace it, but I have no faith in the value to society of patents, and therefore no reason for society (in the form of governments) should play a part in them.
I know somebody (my old Uni Professor) who was a computor himself, back in the "old days".
Well, with the recent moves by Intel and AMD into multiprocessor chips, and the recent news about Pentuim-4 not being much faster than Pentium-III (Pentium-M) because of pipeline issues, I would have thought it clear that yes, the GHz war is on the wane.
Sure, we may have processors in the 4-5GHz range eventually but at the moment, I doubt that things will go a lot further, clock-speed wise. And even a 5HGz CPU is only about 30% faster than the current crop: hardly a big win!
The fact that it can be ported (which I dont disagree with) doesn't mean it would be good to port it for the average user.
Just think how many people now use XBox. It is basically a PC, and yet many gamers went to XBox because of all thr problems they had in getting their PC games to run on their particular hardware.
Platform stability is more important than some people believe.
Interesting that the UK Govt has recently started doing this, albeit in very watered down way. In other projects problems continue to be revealed.
FYI an update to the Callahan story:
Something about DVDs not lasting even 30 years, according to some people. OK, so you copy them to another DVD in 20 years, but how many will slip through the net and not be copied?
I have quite successfully used a Nikon LS30 scanner (8 bit 4000 dpi) on LInux using VueScan. It plugs into the Firewire port using the Kernel "SBP" firewire driver. VueScan does the rest. Colour management is another issue. To get decent results using supplied (approximate) profiles is not good: use a profiling tool such as those sold by Gretag-Macbeth for about £120 ($250?)
Which is it? :-)
Ruth
But online bank accounts ?all? use https -- that is, using SSL as well. Surely that makes it a lot harder to get at useful information on the WEP link?
The thing that bothers me about predictions of "losses" such as these losses in excess of EUR4.5 billion annually due to piracy is that it assumes that the money was available to be paid. In many cases (e.g. children swapping games software) it simply isn't.
The reason is of course that multiplying number pirated * max selling price produced "shocking" numbers for the policy makers, even though it's almost certainly wildly inaccurate.
OTOH, I don't condone software piracy.
Assuming I'm right in believing the poster is from the USA:
I find it very curious that "people of the land of the free" do not trust their govt, and yet do nothing notable about it...
No. Security is designed into (each component of) a product, verified in the implementation, and then verified in the final system. Saying "MS SQL Server is secure" is complete nonsense as a stand-alone statement ... secure against what? in what conditions?
R.
Last time I looked, Europe wasn't "a country". And there are really quite a few broadcasters, including one of the best (good ol'e Beeb!).
I am not doing this, but it is something I would like to do; under the current 'pay for channel bundles' method, most channels piggy-back on the success of a few, while forcing me to 'buy' channels I have no interest in.
I must agree with those who view the posted questions as being a curious choice. Some of my own are: 1. Exactly what is gravity, and can it be controlled ? 2. Can/does anything travel faster than light, with or without wormholes? 3. How can power be generated in ecologically safe ways (that is, without unfortunate byproducts)? 4. How can people fly safely, quickly and affordably without requiring fossil fuels? 5. Is a true 3d hologram (a la Rimmer/Cdr Data/etc) possible?