Tools like sort, uniq, sed, awk and so forth are not "things that almost nobody will use". That is a bit like saying "Who needs coal anyway now we've got electricity?" If you look in your rcscripts sometime, you probably will see plenty of references to these tools.
My Windows installation achieves all the same things my Linux init scripts achieve without needing any of these tools. Also, having a quick look through my init scripts shows me that very few use any of these tools. On my installation,/etc/rc.d/firewall uses both awk and sed, and autofs, i4l and mysql use only sed. Neither sort nor uniq is used at all.
All of these scripts use these tools for purposes that would be better served by writing the script in a language that supported more advanced variable manipulations than the shell. If an equivalent system were used under windows, I would expect the scripts to be written using VBScript or JScript (or, of course, any other language supported by windows scripting host that you have installed), at which point sed and awk would not be necessary.
I could go on, but I suspect we aren't aiming at the same point. My point is that I think it's good to have many small programmes that each do one thing - and do it well - which can then be called from within other programmes. A mail client, for instance, just needs a call to sendmail - it doesn't have to handle the intricacies of SMTP.
The point I'm making is that Windows works in a different way to Unix. The way windows works is that it does have all of these little utility systems, but they aren't programs. Instead, it provides a large number of utility objects in dynamically linkable libraries that can be loaded by programs to achieve the same result. Instead of a shell language like Unix uses, you need something object oriented, like VBS or JS. But the end result is the same, really.
OK, so a lot of objects aren't installed by default. But then the objects aren't generally speaking used by users - they're used by applications, as you suggest. There's nothing to stop the application from installing the object at the same time as the application is installed, which is the way pretty much everything in Windows works.
BTW: the PHP spellcheck functions don't use the unix spell command; they use libaspell. Starting a new process is inefficient compared to using a library.
[note]: OO.o 1.1 contais 1 feature that I find useful that isn't include in office 97, which is PDF export. Office 97 has no features that I find useful that aren't included in OO.o. This is a subjective analysis, but that's the best I can do.
That is fine if you work in seclusion from the rest of the world, but unfortunately MS Office has become an "industry standard". That means that people your business relies on (e.g. clients) often mail you Office documents that they want you to look at and modify. We tried to switch to OpenOffice, but it just didn't cut it - too many incompatabilies. So now we're back to spending $$$ on MS Office licenses for everyone.
In my experience, OO opens about 99% of all MS Word docs just fine. What version did you try? I had problems back in the early stages, and with Star Office 5.2 when I used to use that, but these days I hardly ever have a problem.
Also, I am seeing a wholesale move towards sending PDF documents around, rather than word. My company has a standard line. 'We'll accept your word document, but would prefer PDF because word spreads viruses.' We've managed to persuade quite a few of our clients (and even some suppliers) to switch to PDF that way.
We don't actually use OO on the desktop at the copmany, though. The only reason is that the MD doesn't like the user interface; he prefers the MS Word UI (eugh!).
I don't have any special knowlwge about what MS is doing. But the described approach sound most sensible to me.
And is effectively the way the DRM in Windows Media Player works. It seems likely that they would use a similar system, if not actually a compatible one, for documents.
Perhaps it's my experience with having used Protext on the Amiga, but I never saw the point of an interactive spelling checker. I want to type a document up, get the spelling checked while I'm away doing something else, then come back and correct the misspelt words.
That was great for the days when spell checking was a process that took a long time. But now it can be done almost instantly for any reasonable length document (I have a 35kword document I'm working on with OO.org with a few non-dictionary words in it, and a spell check occurs as quickly as I can move my eyes from clicking the 'skip' button back to the box that displays the word that isn't in the dictionary). The idea of checking the spelling 'while you're away' is an idea that had its time ten years ago, it just isn't possible now. The check will be done before you can get out of your seat.
Windows must have equivalents for spell, sort, uniq and awk, right?
it must have the basics, even on the minimal installation..... surely?
Why do you want things that almost nobody will use in the minimal installation? That would be bloatware, and MS are accused of that enough of the time already.
It's fine that those of us who know enough to be able to use these tools have to go out of our way to get them. It's hardly a difficult task, and well worth the effort.
I've got to take issue with a number of these suggestions:
OUT: 1U (1.75-in. high) servers IN: Blade servers WHY: They save space, eliminate cables and lower costs by sharing power supplies and connectivity.
Don't be ridiculous! Most businesses have no need for more than 1 or 2 servers, for which purpose rack mount is ideal. The server can fit in the same equipment rack as the switches, and in many cases the same rack as the phone extension connections go through as well. Very handy, and probably only costs about 1 - 2,000.
OUT: Color ink-jet printers IN: Color laser printers WHY: Color laser printers used to cost thousands; now they're well under $1,000. And color laser cartridge changes are less frequent--and less messy.
Colour ink-jet. Cost: about 100. Colour laser. Cost: about 700. I know which I'm buying, regardless of maintenance cost. For low volume colour printing (which is all most people are doing at the moment) the inkjet is fine.
OUT: Ethernet hubs IN: Intelligent switches WHY: Newer switches are inexpensive, a prerequisite for IP telephony, and typically support Simple Network Management Protocol for remote manageability.
Hubs are cheaper. Relatively few people use IP telephony. Hubs don't need SNMP because they don't need management. They are 'plug and play' devices in the truest sense of the phrase.
OUT: File servers IN: Network-attached storage appliances WHY: Why maintain file servers for shared storage when you can plug in a simple appliance?
See my other post on this topic. Why pay over the odds for an additional NAS box when the server you already need can perform the same role?
OUT: Serial/parallel ports IN: USB 2.0 ports WHY: The ports won't go away on PCs anytime soon, but for new hardware, Universal Serial Bus peripherals are faster and often easier to set up.
For simple data transfer applications, serial ports are much easier to use than USB. For printing, there is nothing simpler than a parallel port. No need to mess about with USB hubs because you have too many devices - just plug your printer straight into your parallel port. Much easier.
OUT: Token Ring IN: Ethernet WHY: Ethernet: Cheap and ubiquitous. Token Ring: Expensive, with limited vendor sources. Any questions?
Didn't Token Ring go out, like, 10 years ago? I haven't seen anyone attempting to run it since '94, and it was clearly outdated even then...
OUT: File servers IN: Network-attached storage appliances WHY: Why maintain file servers for shared storage when you can plug in a simple appliance?
Uh, correct me if I'm wrong but isn't "network-attached storage appliances" just marketing-speak for "file server"?
There are a few differences. NAS devices are usually use autosensing to configure themselves, typically have integrated backup systems and often come with RAID options for a fraction of the price that most businesses could set up a PC server with the same features (generally due to the lack of server software licensing that is required, most companies wouldn't consider a free server OS). But, yes, essentially it is the same thing.
OTOH, an awfully large number of small companies (that is, the vast majority of the market) are better off with traditional file servers. Why? Because a file server can be a multipurpose machine that also does its time as mail server, web proxy server, accounts server (one of those client-server apps that are dying, I hear you say? err, no, not dying even a little bit, actually), and so on.
A fairly cheap 700 MHz - 1 GHz machine with a relatively small amount of memory should be able to perform all of these functions adequately for a 5 - 50 user network with only intermittent usage requirements (that is, approximately a third of businesses). If Windows isn't used on it, the cost would probably be less than an equivalent NAS box, too.
What the author should have said was that the typical fat-client/server architecture is fading away.
Even that seems wrong to me. My experience suggests that a huge number of such systems are deployed, are being deployed, and will continue to be deployed. This is particularly true in the case of accounts and payroll systems, stock management, and other similar, highly data driven applications. Even web based systems are often managed by client applications these days, rather than web based management. See, for example, Actinic Catalog, one of the leading web shop front ends, particularly in the UK market, or any of numerous content management systems, such as City Desk, which use client side applications for management.
A 'fat' client (not a name I like, I guess it was chosen by proponents of 'thin' clients) is often much more user friendly and intuitive than a thin one, because thin clients are developed for generic situations while fat clients can be customised and include specific features to enhance their usability for their own target application. For instance, web site management software that includes integrated image editors is much easier for the typical user than software that lacks an image editor.
Alas, new DMCAish legal restrictions in the United Kingdom [...]
OK, I've just finished a brief scan of that.
Its rather a big document to receive less than a week's worth of debate before enactment, but I guess I only have my MP to blame for that.
One thing that concerns me substantially is that the phrase "effective technological measure" and a number of similar phrases are used in numerous places in these regulations, but are not defined anywhere. Nor, it would seem, are they defined in any other act of the UK parliament, at least as far back as 1987 (which is as far as HMSO's web copies go). Which means that this law can be readily twisted to mean whatever the first barrister to get his hands on it wants it to mean... yeuch.
Other issues - breaking an anti-circumvention measure is illegal whatever your purpose (unless its for cryptographic research). You could be perfectly entitled to get at the data, but because its protected you can do nothing about it.
The fact is that the copyright directive had an implementation deadline at the end of last year. The UK has just been 10 months late. However, legal precedence in the EU means that until a state has implemented the directive, then it is possible to enforce the directive through the principle of "direct effect".
This means that your material is already - and has been for some time - a copyright violation.
My understanding (I read the directive once, although it's quite a while ago now) is that the directive itself had quite a list of optional exclusions from the anti-circumvention legislation, the only one of which the UK has chosen to enact is the 'cryptographic research' exclusion. However, if a case were tried before the UK enactment (i.e. for events that occur before the end of this month), the full list of exclusions would apply, and these would probably be adequate to ensure that the software in question was legal.
Of course, I ain't no lawyer, so I could be totally and utterly wrong. (Who is it who said that lawyers are always right anyway?)
Unfortunately my 286 is no longer operable. It has a weird proprietory unreplaceable system where the hard drive controller shares an I/O card with the CMOS memory. The hard drive controller is an MFM job, and hence I can no longer get disks that work with it.
A major problem with Kazaa is that it only hashes the first 32K of any file. Any glitches after that go undetected. This is why I won't be sorry to see Kazaa lose out to Overnet, or some other network which hashes the whole file
If I hadn't lost my mod points over the weekend, I'd be modding you up.
If one thing annoys me, its Kazaa zealots who insist that since Kazaa has file hashes, its as unlikely to throw up corrupted files as all the other networks.
Hungarian's hardly forgotten, its still used in a fairly large proportion of Windows development houses.
Machine processable comments are really useful. I've yet to find a more convenient way of writing class level documentation. Python's approach is interesting, and I might favour a similar system, if it were included in a compiled language rather than an interpreted one.
"Compilation into a stack machine form interpreted in software" doesn't really apply to Java, because Java is only stored in stack machine form as an intermediate representation in order to allow the loading mechanism to reason about the safety of the code, it is then translated into native machine code. A variety of implementations can do it in a variety of ways. One interesting approach is translate-on-install, which alleviates almost all of the problems usually associated with this kind of system.
What libraries do C# offer that are not accessible from VB.NET? As far as I know, all C# libraries (at least those in the standard framework) are CLS compliant, and thus accessible from any CLR language.
My understanding is that C# supports pointers, whereas VB.NET doesn't, so there would be some COM objects that could be used from C# but not VB.NET.
Its probably a very small minority of the things people want to do with the system though, particularly seeing as you have to have complete trust to use pointers.
Automated regression tests isn't intended to relieve those lazy programmers, in XP they're the de-facto definition of what the system is designed to. Not to mention that test-first design often leads to better design, in particular wrt coupling between classes and components.
Agreed. It also results in less bugs to start with, because it forces programmers to think about what could go wrong before they start work.
The problem here is that your admin obviously misunderstood the idea of hungarian, which happens a lot. The point isn't to encode the type of the variable (which really is a rather ridiculous idea), but instead to encode information about what you're doing with it.
You can use hungarian sensibly in perl. But I would use it in a substantially different way to the way most Windows API code uses it.
For instance: you could label all information that has been posted from a form with a specific prefix. This would help you determine when you're using a variable whether or not you need to perform security checks on its value first.
Other things that still apply from traditional hungarian would be the 'c' prefix (for a count of number of objects in some category), the 'i' prefix (for indices into arrays). Both of these could be handy.
Hungarian Notation is just file extensions in source code -- it will surprise nobody that both terrible creations were midwifed at the same place.
Uh, you can blame MS for many things, but file extensions were in common usage a long time before they started using them. I think they got the idea from CP/M.
OK, I'll admit to never having heard of these 'HyperQueues' myself, and the docs make it sound like something the developers had come up with themselves.
I think the idea is interesting, its basically a pipelined approach to IPC that means that there don't have to be context switches in order to empty the receiving buffer as often (normally once per message in most messaging APIs). This kind of benefit can only come at some expense, and my guess is that when using the message API for RPC type usage (by far the most common use for messaging APIs, in my experience), it will have no benefit whatsoever, and might even be less efficient. It would be good for streaming applications though, so I would expect to see media handling in this system work quite well.
Microsoft has contemplated this for years as it is a fairly common request. Raymond Chen, whom you might know better as the creator of the wildly popular TweakUI, has been a Windows developer for several years. He has a blog entry describing why they've never done this.
I think to say they've "never done this" is a bit extreme. Word for Windows v1 had a novice mode that hid about half of the menu options. Obviously its not as good an example as that the parent article posted, but it was certainly a start down the correct route.
Also note that many modern MS apps hide menu options that you don't use frequently, which is a kind of adaptive way of doing the same thing. It doesn't work, IMHO, but its certainly thinking along the same ideas...
But yes, I hate the fact that very little software is produced with the idea that people might understand it.
One thing I've come to hate, actually, is one mantra mentioned in the review above, which is "hide the internal model from the user". Why? If I understand the internal model of the application, I can learn how to use it more effectively. If I understand how it stores its data (for example) I can have a quicker grasp of what consqueneces any change will have.
How many people do you know who get totally confused by MS Word's weird behaviours, for instance where deleting the end of a paragraph applies the formatting from the following paragraph to the current one. If people just understood that 'word stores paragraph formatting in a marker just after the end of the paragraph', then it would save a huge amount of confusion. You could have little hints like that in an expert user's version of 'tip of the day'. It would help no end!
What makes bit torrent different from something like Kazaa or Gnutella, just out of curiousity?
BitTorrent was implemented in Python, so everyone around here thinks its kewl;-)
[actually, its just a file transfer mechansim, as opposed to those two which are file transfer + search. The file transfer is *much* better than the above because a substantial amount of effort has gone into optimising the process so that each person uploads copies up to their capacity, and nobody gets overloaded, etc. But that doesn't sound as funny]
If gcc does this, I think it uses 0xDEADBEEF as the value to initialise stuff to, which generally stands out like a sore thumb. But I don't think it does, I think its actually gdb that does that (god knows how, though).
Come to think of it, I'd be surprised if there wasn't a Linux printer driver to do the same thing....
A what?
Linux (and Unix in general) is substantially let down, IMO, by the lack of a clearly defined printer driver interface. Most systems resort to having each program generate a postscript file (through its own integral method) which is then translated by ghostscript to whatever format is required by your printer.
I believe ghostscript does have a PDF output mode (it certainly has a PDF input mode...), but at a guess it isn't very good.
I tend to use pdf[la]tex, which I find gives good quality results easily.
I haven't done this, but have you considered Contracts for Difference? You're trading a derivative, so you'll need to open an account which can be tricky, but from what a friend who has been involved with them in the past has told me it is generally possible to short anything when trading CFDs.
OTOH, the only broker I know of (deal4free.com) doesn't seem to carry SCO.
ObWarning: be careful, you can lose a lot more money than you originally put in when trading derivitives.
Slashdot Mirror
Tools like sort, uniq, sed, awk and so forth are not "things that almost nobody will use". That is a bit like saying "Who needs coal anyway now we've got electricity?" If you look in your rcscripts sometime, you probably will see plenty of references to these tools.
/etc/rc.d/firewall uses both awk and sed, and autofs, i4l and mysql use only sed. Neither sort nor uniq is used at all.
My Windows installation achieves all the same things my Linux init scripts achieve without needing any of these tools. Also, having a quick look through my init scripts shows me that very few use any of these tools. On my installation,
All of these scripts use these tools for purposes that would be better served by writing the script in a language that supported more advanced variable manipulations than the shell. If an equivalent system were used under windows, I would expect the scripts to be written using VBScript or JScript (or, of course, any other language supported by windows scripting host that you have installed), at which point sed and awk would not be necessary.
I could go on, but I suspect we aren't aiming at the same point. My point is that I think it's good to have many small programmes that each do one thing - and do it well - which can then be called from within other programmes. A mail client, for instance, just needs a call to sendmail - it doesn't have to handle the intricacies of SMTP.
The point I'm making is that Windows works in a different way to Unix. The way windows works is that it does have all of these little utility systems, but they aren't programs. Instead, it provides a large number of utility objects in dynamically linkable libraries that can be loaded by programs to achieve the same result. Instead of a shell language like Unix uses, you need something object oriented, like VBS or JS. But the end result is the same, really.
OK, so a lot of objects aren't installed by default. But then the objects aren't generally speaking used by users - they're used by applications, as you suggest. There's nothing to stop the application from installing the object at the same time as the application is installed, which is the way pretty much everything in Windows works.
BTW: the PHP spellcheck functions don't use the unix spell command; they use libaspell. Starting a new process is inefficient compared to using a library.
I don't mean to piss on your parade, but OpenOffice is full of bloat.
I'll buy that. It has a lot of features, and most people won't need most of them. This fits the general definition of bloat.
Some might even say it's more bloated than MSOffice.
I wouldn't go that far, though. The simplest measure of bloat is size / usefulness.
Office 97: Size = 84.7Mb, usefulness = 1, bloat = 84.7
OO.org 1.1: Size = 63.5Mb, usefulness = 1.05 [see note below], bloat = 60.5
[note]: OO.o 1.1 contais 1 feature that I find useful that isn't include in office 97, which is PDF export. Office 97 has no features that I find useful that aren't included in OO.o. This is a subjective analysis, but that's the best I can do.
That is fine if you work in seclusion from the rest of the world, but unfortunately MS Office has become an "industry standard". That means that people your business relies on (e.g. clients) often mail you Office documents that they want you to look at and modify. We tried to switch to OpenOffice, but it just didn't cut it - too many incompatabilies. So now we're back to spending $$$ on MS Office licenses for everyone.
In my experience, OO opens about 99% of all MS Word docs just fine. What version did you try? I had problems back in the early stages, and with Star Office 5.2 when I used to use that, but these days I hardly ever have a problem.
Also, I am seeing a wholesale move towards sending PDF documents around, rather than word. My company has a standard line. 'We'll accept your word document, but would prefer PDF because word spreads viruses.' We've managed to persuade quite a few of our clients (and even some suppliers) to switch to PDF that way.
We don't actually use OO on the desktop at the copmany, though. The only reason is that the MD doesn't like the user interface; he prefers the MS Word UI (eugh!).
I don't have any special knowlwge about what MS is doing. But the described approach sound most sensible to me.
And is effectively the way the DRM in Windows Media Player works. It seems likely that they would use a similar system, if not actually a compatible one, for documents.
Perhaps it's my experience with having used Protext on the Amiga, but I never saw the point of an interactive spelling checker. I want to type a document up, get the spelling checked while I'm away doing something else, then come back and correct the misspelt words.
..... surely?
That was great for the days when spell checking was a process that took a long time. But now it can be done almost instantly for any reasonable length document (I have a 35kword document I'm working on with OO.org with a few non-dictionary words in it, and a spell check occurs as quickly as I can move my eyes from clicking the 'skip' button back to the box that displays the word that isn't in the dictionary). The idea of checking the spelling 'while you're away' is an idea that had its time ten years ago, it just isn't possible now. The check will be done before you can get out of your seat.
Windows must have equivalents for spell, sort, uniq and awk, right?
Yeah. They're right here.
it must have the basics, even on the minimal installation
Why do you want things that almost nobody will use in the minimal installation? That would be bloatware, and MS are accused of that enough of the time already.
It's fine that those of us who know enough to be able to use these tools have to go out of our way to get them. It's hardly a difficult task, and well worth the effort.
I've got to take issue with a number of these suggestions:
OUT: 1U (1.75-in. high) servers
IN: Blade servers
WHY: They save space, eliminate cables and lower costs by sharing power supplies and connectivity.
Don't be ridiculous! Most businesses have no need for more than 1 or 2 servers, for which purpose rack mount is ideal. The server can fit in the same equipment rack as the switches, and in many cases the same rack as the phone extension connections go through as well. Very handy, and probably only costs about 1 - 2,000.
OUT: Color ink-jet printers
IN: Color laser printers WHY: Color laser printers used to cost thousands; now they're well under $1,000. And color laser cartridge changes are less frequent--and less messy.
Colour ink-jet. Cost: about 100.
Colour laser. Cost: about 700.
I know which I'm buying, regardless of maintenance cost. For low volume colour printing (which is all most people are doing at the moment) the inkjet is fine.
OUT: Ethernet hubs
IN: Intelligent switches
WHY: Newer switches are inexpensive, a prerequisite for IP telephony, and typically support Simple Network Management Protocol for remote manageability.
Hubs are cheaper. Relatively few people use IP telephony. Hubs don't need SNMP because they don't need management. They are 'plug and play' devices in the truest sense of the phrase.
OUT: File servers
IN: Network-attached storage appliances
WHY: Why maintain file servers for shared storage when you can plug in a simple appliance?
See my other post on this topic. Why pay over the odds for an additional NAS box when the server you already need can perform the same role?
OUT: Serial/parallel ports
IN: USB 2.0 ports
WHY: The ports won't go away on PCs anytime soon, but for new hardware, Universal Serial Bus peripherals are faster and often easier to set up.
For simple data transfer applications, serial ports are much easier to use than USB. For printing, there is nothing simpler than a parallel port. No need to mess about with USB hubs because you have too many devices - just plug your printer straight into your parallel port. Much easier.
OUT: Token Ring
IN: Ethernet
WHY: Ethernet: Cheap and ubiquitous. Token Ring: Expensive, with limited vendor sources. Any questions?
Didn't Token Ring go out, like, 10 years ago? I haven't seen anyone attempting to run it since '94, and it was clearly outdated even then...
OUT: File servers
IN: Network-attached storage appliances
WHY: Why maintain file servers for shared storage when you can plug in a simple appliance?
Uh, correct me if I'm wrong but isn't "network-attached storage appliances" just marketing-speak for "file server"?
There are a few differences. NAS devices are usually use autosensing to configure themselves, typically have integrated backup systems and often come with RAID options for a fraction of the price that most businesses could set up a PC server with the same features (generally due to the lack of server software licensing that is required, most companies wouldn't consider a free server OS). But, yes, essentially it is the same thing.
OTOH, an awfully large number of small companies (that is, the vast majority of the market) are better off with traditional file servers. Why? Because a file server can be a multipurpose machine that also does its time as mail server, web proxy server, accounts server (one of those client-server apps that are dying, I hear you say? err, no, not dying even a little bit, actually), and so on.
A fairly cheap 700 MHz - 1 GHz machine with a relatively small amount of memory should be able to perform all of these functions adequately for a 5 - 50 user network with only intermittent usage requirements (that is, approximately a third of businesses). If Windows isn't used on it, the cost would probably be less than an equivalent NAS box, too.
What the author should have said was that the typical fat-client/server architecture is fading away.
Even that seems wrong to me. My experience suggests that a huge number of such systems are deployed, are being deployed, and will continue to be deployed. This is particularly true in the case of accounts and payroll systems, stock management, and other similar, highly data driven applications. Even web based systems are often managed by client applications these days, rather than web based management. See, for example, Actinic Catalog, one of the leading web shop front ends, particularly in the UK market, or any of numerous content management systems, such as City Desk, which use client side applications for management.
A 'fat' client (not a name I like, I guess it was chosen by proponents of 'thin' clients) is often much more user friendly and intuitive than a thin one, because thin clients are developed for generic situations while fat clients can be customised and include specific features to enhance their usability for their own target application. For instance, web site management software that includes integrated image editors is much easier for the typical user than software that lacks an image editor.
Alas, new DMCAish legal restrictions in the United Kingdom [...]
OK, I've just finished a brief scan of that.
Its rather a big document to receive less than a week's worth of debate before enactment, but I guess I only have my MP to blame for that.
One thing that concerns me substantially is that the phrase "effective technological measure" and a number of similar phrases are used in numerous places in these regulations, but are not defined anywhere. Nor, it would seem, are they defined in any other act of the UK parliament, at least as far back as 1987 (which is as far as HMSO's web copies go). Which means that this law can be readily twisted to mean whatever the first barrister to get his hands on it wants it to mean... yeuch.
Other issues - breaking an anti-circumvention measure is illegal whatever your purpose (unless its for cryptographic research). You could be perfectly entitled to get at the data, but because its protected you can do nothing about it.
Actually, it has been legal to record a broadcast for private, non-commercial use since 1988.
IANAL, etc.
The fact is that the copyright directive had an implementation deadline at the end of last year. The UK has just been 10 months late. However, legal precedence in the EU means that until a state has implemented the directive, then it is possible to enforce the directive through the principle of "direct effect".
This means that your material is already - and has been for some time - a copyright violation.
My understanding (I read the directive once, although it's quite a while ago now) is that the directive itself had quite a list of optional exclusions from the anti-circumvention legislation, the only one of which the UK has chosen to enact is the 'cryptographic research' exclusion. However, if a case were tried before the UK enactment (i.e. for events that occur before the end of this month), the full list of exclusions would apply, and these would probably be adequate to ensure that the software in question was legal.
Of course, I ain't no lawyer, so I could be totally and utterly wrong. (Who is it who said that lawyers are always right anyway?)
Unfortunately my 286 is no longer operable. It has a weird proprietory unreplaceable system where the hard drive controller shares an I/O card with the CMOS memory. The hard drive controller is an MFM job, and hence I can no longer get disks that work with it.
:-)
It makes a cool wall ornament, though
A major problem with Kazaa is that it only hashes the first 32K of any file. Any glitches after that go undetected. This is why I won't be sorry to see Kazaa lose out to Overnet, or some other network which hashes the whole file
If I hadn't lost my mod points over the weekend, I'd be modding you up.
If one thing annoys me, its Kazaa zealots who insist that since Kazaa has file hashes, its as unlikely to throw up corrupted files as all the other networks.
Yeah, right.
I gotta take issue with a gewof your points.
Hungarian's hardly forgotten, its still used in a fairly large proportion of Windows development houses.
Machine processable comments are really useful. I've yet to find a more convenient way of writing class level documentation. Python's approach is interesting, and I might favour a similar system, if it were included in a compiled language rather than an interpreted one.
"Compilation into a stack machine form interpreted in software" doesn't really apply to Java, because Java is only stored in stack machine form as an intermediate representation in order to allow the loading mechanism to reason about the safety of the code, it is then translated into native machine code. A variety of implementations can do it in a variety of ways. One interesting approach is translate-on-install, which alleviates almost all of the problems usually associated with this kind of system.
What libraries do C# offer that are not accessible from VB.NET? As far as I know, all C# libraries (at least those in the standard framework) are CLS compliant, and thus accessible from any CLR language.
My understanding is that C# supports pointers, whereas VB.NET doesn't, so there would be some COM objects that could be used from C# but not VB.NET.
Its probably a very small minority of the things people want to do with the system though, particularly seeing as you have to have complete trust to use pointers.
Automated regression tests isn't intended to relieve those lazy programmers, in XP they're the de-facto definition of what the system is designed to. Not to mention that test-first design often leads to better design, in particular wrt coupling between classes and components.
Agreed. It also results in less bugs to start with, because it forces programmers to think about what could go wrong before they start work.
The problem here is that your admin obviously misunderstood the idea of hungarian, which happens a lot. The point isn't to encode the type of the variable (which really is a rather ridiculous idea), but instead to encode information about what you're doing with it.
You can use hungarian sensibly in perl. But I would use it in a substantially different way to the way most Windows API code uses it.
For instance: you could label all information that has been posted from a form with a specific prefix. This would help you determine when you're using a variable whether or not you need to perform security checks on its value first.
Other things that still apply from traditional hungarian would be the 'c' prefix (for a count of number of objects in some category), the 'i' prefix (for indices into arrays). Both of these could be handy.
Hungarian Notation is just file extensions in source code -- it will surprise nobody that both terrible creations were midwifed at the same place.
Uh, you can blame MS for many things, but file extensions were in common usage a long time before they started using them. I think they got the idea from CP/M.
OK, I'll admit to never having heard of these 'HyperQueues' myself, and the docs make it sound like something the developers had come up with themselves.
I think the idea is interesting, its basically a pipelined approach to IPC that means that there don't have to be context switches in order to empty the receiving buffer as often (normally once per message in most messaging APIs). This kind of benefit can only come at some expense, and my guess is that when using the message API for RPC type usage (by far the most common use for messaging APIs, in my experience), it will have no benefit whatsoever, and might even be less efficient. It would be good for streaming applications though, so I would expect to see media handling in this system work quite well.
Microsoft has contemplated this for years as it is a fairly common request. Raymond Chen, whom you might know better as the creator of the wildly popular TweakUI, has been a Windows developer for several years. He has a blog entry describing why they've never done this.
I think to say they've "never done this" is a bit extreme. Word for Windows v1 had a novice mode that hid about half of the menu options. Obviously its not as good an example as that the parent article posted, but it was certainly a start down the correct route.
Also note that many modern MS apps hide menu options that you don't use frequently, which is a kind of adaptive way of doing the same thing. It doesn't work, IMHO, but its certainly thinking along the same ideas...
But yes, I hate the fact that very little software is produced with the idea that people might understand it.
One thing I've come to hate, actually, is one mantra mentioned in the review above, which is "hide the internal model from the user". Why? If I understand the internal model of the application, I can learn how to use it more effectively. If I understand how it stores its data (for example) I can have a quicker grasp of what consqueneces any change will have.
How many people do you know who get totally confused by MS Word's weird behaviours, for instance where deleting the end of a paragraph applies the formatting from the following paragraph to the current one. If people just understood that 'word stores paragraph formatting in a marker just after the end of the paragraph', then it would save a huge amount of confusion. You could have little hints like that in an expert user's version of 'tip of the day'. It would help no end!
What makes bit torrent different from something like Kazaa or Gnutella, just out of curiousity?
;-)
BitTorrent was implemented in Python, so everyone around here thinks its kewl
[actually, its just a file transfer mechansim, as opposed to those two which are file transfer + search. The file transfer is *much* better than the above because a substantial amount of effort has gone into optimising the process so that each person uploads copies up to their capacity, and nobody gets overloaded, etc. But that doesn't sound as funny]
If gcc does this, I think it uses 0xDEADBEEF as the value to initialise stuff to, which generally stands out like a sore thumb. But I don't think it does, I think its actually gdb that does that (god knows how, though).
Come to think of it, I'd be surprised if there wasn't a Linux printer driver to do the same thing....
A what?
Linux (and Unix in general) is substantially let down, IMO, by the lack of a clearly defined printer driver interface. Most systems resort to having each program generate a postscript file (through its own integral method) which is then translated by ghostscript to whatever format is required by your printer.
I believe ghostscript does have a PDF output mode (it certainly has a PDF input mode...), but at a guess it isn't very good.
I tend to use pdf[la]tex, which I find gives good quality results easily.
I haven't done this, but have you considered Contracts for Difference? You're trading a derivative, so you'll need to open an account which can be tricky, but from what a friend who has been involved with them in the past has told me it is generally possible to short anything when trading CFDs.
OTOH, the only broker I know of (deal4free.com) doesn't seem to carry SCO.
ObWarning: be careful, you can lose a lot more money than you originally put in when trading derivitives.
Will people already stop fucking using wikipedia as some sort of definitive authority on every subject under the sun!
Not given current trends, no.
POST checks are for wimps! ;-)