The idea is not as idiotic as some might suggest. Most of them are based on EXISTING technology.
No-one's suggesting that MS has to fill in forms to issue products, but that the base Windows product is made available after it meets standards.
Linux runs a stable base and development code with their even and odd kernel releases. Features of the odd kernel are integrated into the even kernel only on some sort of committee say so. [This is what inspired the solution, if you must know] The main difference here is that we're dealing with a company with a past history of disrupting the market.
The thing does not even ban MS from improving the kernel or any other feature. But this improvement is to be made as an add-on, just as they have in the past, and they have to make the system without the addon available. What it does prevent is MS fostering IE onto everyone as a requirement to have the OS. I mean, MS has the installation down to two clicks, if they are to be believed.
The idea that some feature of the driver might be made available to all operating systems is inspired by Microsoft's microdriver model.
The idea of a bios-level install is inspired by products like OS/2's Boot Manager, Compaq's and IBM disk based BIOS utilities, partition magic and others. The idea is that one could use a generic install for OS systems, and then have the OS compile its own drivers from the stored microdrivers. The thing can be run out of its own partition, like boot manager and Compaq's bios utility are.
There is nothing in the `BIOS level' system install and maintanence that prevents an OS from having a different file system, loading additional drivers, including those on demand or whatever. All that is required is basic installation and a reasonable support for the principle features of the computer.
The idea of being able to roll back the system is to prevent some sort of `not available' noise. I mean, you can have Windows and a number of service packs on the same cdrom, and the installation could pick files from the base product or the base product and service packs, based on selecting a different install file.
The idea of retrofitting older OS installs into the picture is to help undo the damage done by MS anti-competitive actions, much as AT&T had to undo their wall-plugs.
Maybe some of the presentation was a little too extreme for some to follow, but the heart was there.:)
When you come to grips with the concept that the home grown data and programs are more valuable to the users than the manufactured software, then you will come to understand why retro support is so essential. We had some weird GWBASIC program running right up to 1999, because the data it processed was more valuable than replacing it with something easier to program.
The sort of solution protects Microsoft's IP rights and their right to innovate, etc, but it also protects the consumer from these very things. We don't stop MS from making Outlook, but we prevent them from forcing it onto everyone as a condition of Windows.
Another solution would be to require the availability of older versions. I mean, if Win98 is that much better than Win95, then let the public speak.
The truth is that when a certian share of the market is had, then some sort of regulation should kick in. The situation in changing the Windows UI should be no different to changing the position of controls on a motor car.
I suggest that Windows be frozen at certian levels, and upgraded by add-on patches. Periodically, these patches will be examined by some external committee, and those deemed essential will be made into a standard patch, to be distributed with all copies of Windows.
Users would then be able to apply the OS or the OS with the standard patch level. The level of patching will be such that any of the last x versions, or the last x years, might be restored from a given copy. That is, if I buy Win2002, I should be able to roll out the standard Win95, Win98, WinME, &c.
New features would be distributed under separate cover. These will be greatly restricted, because they may only be added under the result of an external committee.
The Peter principle does cover this. It's mentioned in the first book on the subject.
While Peter's book says that people rise to the level of incompetence, he also acknowledges that certian people on both sides will appear to buck the norm, and lists them.
Scott Adams [Dilbert] is more drawn on the management's attempts to cope with their reality not coinciding with the underlying actualities. The main source of the friction is that mgt does not seem to understand that efficieny != effectiveness, and it is effectiveness, not efficiency, that keeps customers happy.
In one case, the equation of effectiveness (which can not be measured) = efficiency (which can) produces humourous results. The sad truth is that one turns an organisation aimed at getting results for the clients [effectiveness] into one that can produce more blue tokens than any other [measurable efficiencies].
The other case is that unlike the fire triangle (heat + feul + air) which is to be broken, the efficiency triangle (faster+better+cheaper) is to be assembled out of incompatable parts and is rarely done. Management assumes a soulution exists in all cases, and engineers know a solution exists in rare cases. This friction generates some of Dilbert's humour.
How to get open standards onto computers. This could be made to work. I mean, AT&T had to undo their phone-jack thing. The real issue is not in the vendor-issued stuff, but in the applications and data written for specific OS's are worth a lot more than the OS themselves. This is why retrofitting is very important
Require new PCs to be dual booting, with a required different companies to provide the OS.
This could be done even further by supplying some sort of BIOS level OS install and driver manager.
So for example, you install in some boot-manager partition, the drivers for the attached hardware, and some boot manager utility. You then boot into this, run fdisk from here, and install the OS, which would configure its drivers from data stored in the OS and in the boot-partition.
Require this to be retrofitted for OS's written in the last 10 years or so. This could mean that some sort of OS installer/driver library thing would be written and distributed at cost. You could then run this installer on a 486, and then install a OS/2 or linux cd and install these on the 486. The installer would then build generic drivers for the OS in question on the fly. The OS installer could then boot from some partition and copy things from the original cd and the partition.
The base code of MS Windows to be approved before release, issued frozen, or otherwise issued that a predefined level is issued. The binaries can be made available until user references ceases (something like air-craft components?). For example, Windows 3.x + DOS is still a good client for running under BeOS, Linux, OS/2 or Win32. Vendors can then add certian features on top of the base code, sort of like how Linux-shells work. If MS wants to make improvements, these can be done via plugins, like they have done in the past [Multimedia Extentions for Win30, Win32 extenstion for Win31, IE for Win95, Media Player]. Something like how Apple apprently uses a BSD base code, I'm told.
Remove all of the fluff from the base Windows OS. For example, the WebTV and IE stuff could be removed, because these were not always in Win32 OSes, and are not needed.
Actively encourage emulation level Win32 and Win16 clients for other OS's. In the first level, MS would be required to actively assist in the completion of Win16 clients for a wide range of OS's, like Linux, OS/2, BeOS and Apple. In the second round, a Win32 client with features compatible to Win95.
Actively deny MS access to certian markets, in order to promote open market. For example, It might be set up to deny MS access to key internet features. (users would have to use the alternate OS for these). MS could be denied first-mover rights on Windows for key issues.
Requirement to use some open format document standard. This would allow for alternate companies to access documents, much as we do now with bitmaps, movies, data bases, web pages, etc.
Unless we do something like this, the whole ecconomy can be held hostage by a single company.
os2fan is aware of what the jargonfile says of OS/2. The jargonfile is available as an OS/2 INF file. It refers to version 1.x, mainly. Version 2.x and later is considerably better.
The same could be said of any of the OSes in the first five years of existance. Windows, that lame dog of an OS from 1984, did not mushroom until after vers 3.x, and Linux did not go anywhere until about 5 years or so later.
I know OS/2 has some bad things about it, but in the main, its quite usable. When you run allocmem it gets rid of the unnecessary dll files that load (specifically, pages them out to swapper.dat). This gives you heaps of core for the apps.
Interesting note: {flamebait}Windows NT derives from v 1.3 of OS/2, and its still there in W2K. I mean, the IBM OS/2 1.3 cmd.exe and rexx runs natively under it. [I've done this, based on a thing in Technet.]{/flamebait}:)
The real reason for wanting an OS/2 laptop is that we can force open standards and a universal installer. For everyone.
If you read the article, you find a different question to what really is happening.
The reindeer do not fly. If the flight is coming from the reindeer, then the sleigh would hang. Since it does not hang, the reindeer are normal reindeer, and the whole is set on an invisible road.
Time zones and presumably, the large number of households do not matter. Santa is shown as being back at home on christmas morning. The explination is that santa, like OS/2, is a pre-emptive multitasker.
Once you understand this, then the whole article becomes some sort of skill at showing how the author can do calculations. It's not really good science, since it does not open the consideration that the propositions of the calculations are wrong.
os2fan believes that santa is real (for all that's worth)
I have used base 120 since the seventies, using only 12 digits. You don't need 120 digits. You do not have to learn lots of tables either. I use only as far as 12*12 tables.
In practice, you use alternating columns of 10 and 12, eg 2001 = 1681 = (((1*10+6)*12+8*10)+1). The number is written in any form that makes the parity of the column clear. This means either groups of 2, 4 or 6 digits, or with dots and commas.
The notation I use is V and E for the 10 and 11 digits, and commas and dots to separate different orders of numbers to make them more readable, eg
2**64 = 3.69.01.57.31.E9.50.07.90.16. This can be written as 369.0157.31E9.5007.9016, or as 369,0157.31E9,5007.9016.
The radix is a colon, eg pi=3:16E8.E3....
The number is read using V0 as TEENty, and E0 as eLEFTy, as tenty and eleventy are confusable with twenty and seventy. A hundred is 1.00, a thousand is 1.0000.
Fractions are always read in pairs, not singly, thus pi = three point sixteen, eleftyeight, eleftythree,...
Zeros in the number are dealt with as follows.
If a zero at the end of a decade is needed, then it is neat, eg 50 is fifty or fiftyneat. A zero at the end of a number is required. Thus, one and a half is 1:60, not 1:6.
If a zero is needed before in the tens column, then is is oh, eg one - oh-eight, for 1.08. The zero at the start of a number can be suppressed in writing and speaking: thus 1.08, not 01.08.
In a semimedial zero, at least the oh-form is given. Eg 50.07 is fifty-oh-seven, or fiftyneat, oh-seven, not fifty, seven (which could be 57). The 0.0 is a semimedial zero.
A special trailing zero is used to force the number into reverse format. This makes each one count as 10, and the order of counting is in by 12s first, ie in dozens. This is the d symbol, eg 12 doz and 5 is 12.5d It is rarely left in this form, but essential.
You can see from even this smattering that alternating bases are radically different to the simple bases like 10 and 12.
As an aside, one of the multiplication rules and a certian Linux-friendly penguin share the same name.
Best to select a composite number bigger than 365, and then leave days out. For those who peddle binary calendars, here's a data structure that I use in all forms of BASIC.
Since the year has 12 months, and no month has more than 31 days, you can store a date into a number in the form N = Y*372+M*31+D-32-C, where C is a constant. You can store 89 years into a single signed number, and if you fiddle with negatives, you can get 178 yr, 2 month into a 16-bit number.
If we put packed_date=Y*365+M*31+D-32-C, then we can get Y=(packed_date+C)mod 365, etc. This scheme works because each date has a unique number, although not all numbers correspond to days.
Another trick for the calendar is to start the year on 1 Mar, making the leap year the last day of the year. This makes the length of the month equal to 31-((M-1)mod 5 mod 2).
Also, any proposal that puts days outside the month, eg New year's day not in any month, etc are going to run afoul of the programs that expect days to be in months.
I have been using base 120 since 1977 (dec) *Serious*. I've used and actively pushed other number systems as well,(8 10 12 14 16 18 20 22 28 30 36 40 42 60 80 120 1680, if you must know:) )so, I'm quite familiar with the benefits of these!!!
Base 120 beats them all hands down. Not only do all of the numbers 1/(2**a*3**b*5**c) terminate, but 7, 11, 17 have refreshingly short periods. eg 1/7 is 0:17.17.17...,
If you want something smaller, try, eg 18 [where 7*7*7=111, and so 7 divides 1/7, and the same is true for 5 dividing 1/5].
Those of you who think it's a hard base to calculate in have never tried. It's actually easier to work in than base 60.
Historically, it has been used by the Germanic nations, the conversion taking place around 900 onwards. Check out Wright's `Old Norse', where the numbers are given, like "200=one hundred and eighty".
I've seen all of this before. This may appear off topic, but we have to remember that the numbers describe our perceptions of things. For example, the form of 73 is such that we might think it belongs to a group of numbers 70-79, and named after 70 (ie 73 is in the seventies). This can affect how we see how we fit in the longness of time.
One thing that is easily forgotten is that cardinal time systems are backwards-looking, where ordinals are forward-looking.
In a cardinal system, the elapsed time is given, eg 2000.0000 years. But the point this starts is at the very start of the era, and any point in that 2xxx.xxxx refers to 2000.0000.
In an ordinal system, the elements are named on the part completed one, and so the ends of the era are named, that is, a future point is named. So when we refer to `the third millenium', we refer to `3000', and all dates from the previous millenium point to this, ie '2001-3000'.
The issue here is not so much that `the third millenium starts on 1 jan', but that `if we use a cardinal system, we are harking for the dark past, not the deep future'.
My view on the millenium start is that centuries and higher should carry ordinal names (20 th c, 21th C, 3rd mill), whereas decades and lesser carry ordinal names (eg 1970s, 1963). Moreover, the centuries and millenia align with the decade-starts (so that 21th c runs from 2000.000 to 2099.999).
As to calendar reform, the system is not all that difficult to deal with, and the added variety of the movable feasts and so forth adds a measure of unpredictibility to the calendar. Over here, the late Easter of 2000 ran straight into Anzac day, so we had a five day weekend:):):)
There is considerably more to the changing of the calendar than the measurements, since people more often observe the passage of similar numbers in the calendar, than in the linear system. (You do not observe passages of 1 mile or 100 miles after anywhere as often as 1 year or 1 month after).
This is not likely to succeed, for the following reasons:
Calendars are deeply tied to the religion, an attack on the calendar is seen as an attack on the religion, as the French discovered in 1790
A fixed week calendar is grossly unfair especially if your birthday falls on a tuesday, and someone else's falls on a sunday each year.
Investment in the current calendar is high so it is not likely to succeed, because of the associated costs.
None the same, a transitional calendar can be made to work, as long as the two are seen to co-operate. The calendar can expanded in operation, or left as a passing fad. We use base 16 in computers in much the same way:)
The planetary names of the week do follow a logical order, though the logic is more obscure today. The seven day week can be made into 10 days by adding days as follows: Monday, Uranus, Tuesday, Wendneday, Neptune, Thursday, Friday, Pluto, Saturday. Unravel these using *7 mod 10, and you will see the Greek order of planets, with the three modern ones at the end.
Clock reform is more closely allied to the weights and measurements. The measurement system works as well under the Christian or Moslem calendars, but fail under a decimal system, because the day is a fixed measure, and measurement systems rely on the second, not the month.
The native metric system is to divide the day into 40 kc , and then into 1000 c, and so forth. Then, the km/kc = m/c. Metric + Decimal Days = unusable, as the km/h is too slow, and the m/s is way too fast.
If you intend to use a hundred-based system (day = 10 hrs of 100 min of 100 sec), the measurement system should be 100-based (eg mile of 100 ch of 100 ft of 100 lines).
A good number of users, even power users, are not programs. Many people are interested in using the available tools than writing new fancy ones. In order for some people to be effective programmers, power users, etc, they need pre-existing toys to improve and adapt.
Of course, once you invest a large amount of time and effort into databases and scripts that rely on the quirks of some tool (either commercial or open source), you are not keen on tossing the whole lot out for new and better tools. Especially if the new and better tools do not fit into the hole the old one did.
Many of us just like to unwrap the present and play with it, not spend the half-afternoon making the toy and then seeing if it fits our needs. Even if the compiled binary were 85% effective, it will give us an idea of what it does. {sarcasm}The largest software manufacturer in the world makes 85% effective software, and you feed in the other 85%!{/sarcasm}
I thought there were surveys around since the world war, that says this in effect:
"If you make changes to people's environment, and test the results, then you will see a plus change in productivity. This is true if you add a change, and then take it." Kind of like being in the spotlight...
It would have been interesting to see what the increased productivity was: more lines of code?
OK, so people can rate their own pages. So why are we leaving it to the big end of town to tell us what we can watch, if we want rated pages...
I mean, it's all well and fine to have rated pages. We need to take the next step, which should be like this:
Imagine a browser ratings engine that works off a standard list. You download a rating-profile from your site. It says `allow x, deny y'. Were the ratings-table a standard format, then you can download for any browser, and any society, a rating table. For example, you could use Netscape and a AntiScientology page, or a `no evolution' page. You should be able to load multiple issues, eg no-evolution + no-commie-stufff.
Maybe we could edit our own as well...
The thing is, if it's there and not working, it might as well not work. If we want to make it work, then we need to push the UI somewhat.
NO! NO! NO! we don't want GREP either...! If we can't come up with a better use of computers than GREP, than who can...
If you look at some hotel guide, you see all of these silly little icons, like `food', `showers' etc. You can work out what facilities are available from these.
Now imagine something like a search engine, picking up these ratings icons, and filtering them according to requests. You want no {x}, we'll screen out the {x} or grey them out or whatever your preference is.
Site rating would be done at the first at peer level. There would be some rights of appeal, to a certian number of authorities. There would be about three or four levels of grievence appeals each costing additional money.
Sites can decline to participate, and you can decline to use the ratingware on your system. But in both cases, it's like going outside. You don't know what sort of strangers you will meet...
Sites should also be able to reject an applied rating (as in opt out), even to the extent of posting a `don't disturb' sign on the door.
Say I create a site "occulthelp.org". It may get rated as some setting, eg "occult" by things that look for sting. On the other hand, the site might be a genuine resource for families that are recovering from excessive occultdom in one of the children.
If the site were subjected to peer review, the rating might reflect the true help-recover-from nature of that site, rather than a `this is more'.
You could have three or four or so chanels of censorship, by different groups. You as a user might then select the ratings engine as you select the search engine.
I mean, if we want anything better than big business, grep or whatever looking after our ratings, we have to do something ourselves on it.
Hence this suggestion...:)
It sounds pretty optional, but it is a lot better than name-guessing, grepping and whatever. I mean, if you as a page opt out, you can not be seen by those who have ratingware running. It's your call, not someone else's
OK, so the international scene is not going to be easy to get uniform standards. But there's a better than allowing programs to block any site that has the word `extreme' in it.
Peope are looking for some sort of content control... Either
we're going to leave it to the big end of town to decide what we watch, or
netizens are going to police the rating scale, or
we're going to run a zillion standards that people are going to make uniform, or
[Heaven forbid] they might kill off the net...
In any case, if we don't make a move and get some sort of ratings systems, (any ratings systems), we are gong to live of grep's rating.
Maybe they could rate sites like they rate movies. Then you could visit sites based on their ratings. A browser could be set to browse only rated sites for example, and stiff fines applied for persons who abuse the rating system (eg turning a site rated as general exhibition into a porn site, for example)
Let us not forget that a UI is a communication chanel to the user. Many of the later shells have become excursions in what is technically possible, and forget the user.
An interface should give the user some indication on what to do, and somewhere to do it. The basic function of what to do could be explained in the manuals, like mouse clicks, etc. But the rest of it should be visible.
For example, the CLI emulates a teletype, with a few standard hacks. The CLI brings to the TTY, the notion of redirection of output, a programming interface, and pipelineing. Once mastered, the CLI can prove to be a powerful workhorse.
In a windowing system, you can run multiple CLI sessions. You would be very upset if commands made in one session affected another. This is called leaking. Yet, different command lines could be logged to the same log file, what is going on would be clear if different names are used: consider the IRC sessions, for example.
What makes GUIs scary is that it is all too easy to hide things, and move them from place to place. The interface varies from user to user, and from vendor to vendor, and from version to version.
A small UI, like the Win3x shell, makes a dandy thing if the main activities is to launch intensive applications, like games, or to load different users, or to install hardware and software to the OS. The shell is small, and fits nicely into ram, and is more userfriendly than a command line. A fixed shell makes it easier to describe in manuals, for one-time tasks like installation, or rare activities like fs maintainance.
Once into an activity, you can then load the shell of your choice (on a user by user basis), and your customisations for that user. This can be a full set of widgets and hacks designed to allow programs to integrate.
What happens now is that each new version of Windows seems to say `lets make Windows look like OS/2 v 1 [so we got Program Manager and File Manager]. Let's make it look like v 2.0, so we get Explorer. Let's make it look like Netscape, so we get the Win98 explorer thing. And let's make it look like the Adobe interface, so we get this Win.NET thing.
So if you want to do any nice things, you have to learn the weird languages that make the different shells tick, or just leave them alone.
And so you see why, despite the power of the shells, most people use them in much the same way as the dodgy little menus that came with the DOS machines in the late 80's. Use the shell to start programs. All the rest of the shell is so much lard.
The point is that the technology exists, and that ASIO (or FBI or whatever) are not the only ones who may want to snoop.
We are seeing the leading edge of what is now large scale computer snooping, eg Carnivore.
Eventually, a lot of crunching external data may become legit, like SETI@home is. It's not that far off that we'll start seeing `commercial in confidence' snooping.
Also, you don't need to nail the person on the collected data, all you need to do is alert that this person is worth watching, and nail on some legal-to-collect thing.
And being companies, there is little around the commercial in confidence stuff, and even less if there is a one-way ID system. [eg file `os2fan' contains no references to file `votegeek', but file `votegeek' contains commentry on `os2fan'.
The wake-up call is not that the spooks are trying it, but that it can be done. Since the law enforcement officers has to convince the judge that what they did is right, there is this protection from then.
Much of the command line interface already exists in the teletype machines. In fact, you talked to computers via teletype machines. Batch processing preceded teletype machines, because the punched card and batch runs were cheaper to lay on then terminals (either glass ttys or the real thing) for most users.
Think of a computer connected to a tty machine. You have a program looking after the console. It reads input, it runs output.
You can build in commands to it.
You can endower a simple or complex macro language to it.
You can allow it to store output to other devices, or files.
You can allow it to gobble up bits of the output as input.
Sounds like a command shell to me, like 4os2, command.com or csh or bash. These differ in what they can do, how good is their macro language, or the built inservices, but they are much the same otherwise.
The UI has been around a long time before computers, and guess what? - it works. It even works under a GUI.
And, you will find that users would get terribly upset if the pipe or keystrokes from one tty session bled through to another...
So yes, in the beginning there was a lonely command line. But the command line needed to talk to a something. At first it phoned up other command lines, and talked to them. So was born the telegram, the telex, and the teletype.
May years passed, and they needed to talk to computers. Computers could not talk, but they could type, and who better than a telex to talk to. So we started by telexing the computer.
Telexes were expensive, and most users had to use punched cards. But when the teletype terminals became cheaper, the punched cards disappeared, and so on.
Terminals became more capable, and one could deal with the terminal as a changing page, rather than a tty session. You could use the screen as a window on a page, and from this, via a long and winding road, we get character based terminal sessions, windows, and finally, the ultimate in perfection for the GUI, the workplace shell...
The real issue is that the electronic copy of a song, be it an MP3 or wave file, is somewhere between a copy and a broadcast. Artists should be reimbursed for some sort broadcast/copy exposure, but how? And, quite frankly, we're not going to get peace until we pay our dues.
I really don't know the answer myself, but I'll toss in some thoughts. Consider for example, some sort of pool of money collected from the distribution of media. As you buy the media, you acquire a certian licence to hold a volume of MP3s. Artists would be paid some sort of royalty based on much the same way as radio royalties, with some additional basis for the `underground' stations.
I know it may hurt, but I'm afraid that if the copy process can not be trusted, then that's the way things are going to head.
Slashdot Mirror
No-one's suggesting that MS has to fill in forms to issue products, but that the base Windows product is made available after it meets standards.
Linux runs a stable base and development code with their even and odd kernel releases. Features of the odd kernel are integrated into the even kernel only on some sort of committee say so. [This is what inspired the solution, if you must know] The main difference here is that we're dealing with a company with a past history of disrupting the market.
The thing does not even ban MS from improving the kernel or any other feature. But this improvement is to be made as an add-on, just as they have in the past, and they have to make the system without the addon available. What it does prevent is MS fostering IE onto everyone as a requirement to have the OS. I mean, MS has the installation down to two clicks, if they are to be believed.
The idea that some feature of the driver might be made available to all operating systems is inspired by Microsoft's microdriver model.
The idea of a bios-level install is inspired by products like OS/2's Boot Manager, Compaq's and IBM disk based BIOS utilities, partition magic and others. The idea is that one could use a generic install for OS systems, and then have the OS compile its own drivers from the stored microdrivers. The thing can be run out of its own partition, like boot manager and Compaq's bios utility are.
There is nothing in the `BIOS level' system install and maintanence that prevents an OS from having a different file system, loading additional drivers, including those on demand or whatever. All that is required is basic installation and a reasonable support for the principle features of the computer.
The idea of being able to roll back the system is to prevent some sort of `not available' noise. I mean, you can have Windows and a number of service packs on the same cdrom, and the installation could pick files from the base product or the base product and service packs, based on selecting a different install file.
The idea of retrofitting older OS installs into the picture is to help undo the damage done by MS anti-competitive actions, much as AT&T had to undo their wall-plugs.
Maybe some of the presentation was a little too extreme for some to follow, but the heart was there. :)
When you come to grips with the concept that the home grown data and programs are more valuable to the users than the manufactured software, then you will come to understand why retro support is so essential. We had some weird GWBASIC program running right up to 1999, because the data it processed was more valuable than replacing it with something easier to program.
The sort of solution protects Microsoft's IP rights and their right to innovate, etc, but it also protects the consumer from these very things. We don't stop MS from making Outlook, but we prevent them from forcing it onto everyone as a condition of Windows.
The truth is that when a certian share of the market is had, then some sort of regulation should kick in. The situation in changing the Windows UI should be no different to changing the position of controls on a motor car.
I suggest that Windows be frozen at certian levels, and upgraded by add-on patches. Periodically, these patches will be examined by some external committee, and those deemed essential will be made into a standard patch, to be distributed with all copies of Windows.
Users would then be able to apply the OS or the OS with the standard patch level. The level of patching will be such that any of the last x versions, or the last x years, might be restored from a given copy. That is, if I buy Win2002, I should be able to roll out the standard Win95, Win98, WinME, &c.
New features would be distributed under separate cover. These will be greatly restricted, because they may only be added under the result of an external committee.
While Peter's book says that people rise to the level of incompetence, he also acknowledges that certian people on both sides will appear to buck the norm, and lists them.
Scott Adams [Dilbert] is more drawn on the management's attempts to cope with their reality not coinciding with the underlying actualities. The main source of the friction is that mgt does not seem to understand that efficieny != effectiveness, and it is effectiveness, not efficiency, that keeps customers happy.
In one case, the equation of effectiveness (which can not be measured) = efficiency (which can) produces humourous results. The sad truth is that one turns an organisation aimed at getting results for the clients [effectiveness] into one that can produce more blue tokens than any other [measurable efficiencies].
The other case is that unlike the fire triangle (heat + feul + air) which is to be broken, the efficiency triangle (faster+better+cheaper) is to be assembled out of incompatable parts and is rarely done. Management assumes a soulution exists in all cases, and engineers know a solution exists in rare cases. This friction generates some of Dilbert's humour.
- Require new PCs to be dual booting, with a required different companies to provide the OS.
This could be done even further by supplying some sort of BIOS level OS install and driver manager.
So for example, you install in some boot-manager partition, the drivers for the attached hardware, and some boot manager utility. You then boot into this, run fdisk from here, and install the OS, which would configure its drivers from data stored in the OS and in the boot-partition.
- Require this to be retrofitted for OS's written in the last 10 years or so. This could mean that some sort of OS installer/driver library thing would be written and distributed at cost. You could then run this installer on a 486, and then install a OS/2 or linux cd and install these on the 486. The installer would then build generic drivers for the OS in question on the fly. The OS installer could then boot from some partition and copy things from the original cd and the partition.
- The base code of MS Windows to be approved before release, issued frozen, or otherwise issued that a predefined level is issued. The binaries can be made available until user references ceases (something like air-craft components?). For example, Windows 3.x + DOS is still a good client for running under BeOS, Linux, OS/2 or Win32. Vendors can then add certian features on top of the base code, sort of like how Linux-shells work. If MS wants to make improvements, these can be done via plugins, like they have done in the past [Multimedia Extentions for Win30, Win32 extenstion for Win31, IE for Win95, Media Player]. Something like how Apple apprently uses a BSD base code, I'm told.
- Remove all of the fluff from the base Windows OS. For example, the WebTV and IE stuff could be removed, because these were not always in Win32 OSes, and are not needed.
- Actively encourage emulation level Win32 and Win16 clients for other OS's. In the first level, MS would be required to actively assist in the completion of Win16 clients for a wide range of OS's, like Linux, OS/2, BeOS and Apple. In the second round, a Win32 client with features compatible to Win95.
- Actively deny MS access to certian markets, in order to promote open market. For example, It might be set up to deny MS access to key internet features. (users would have to use the alternate OS for these). MS could be denied first-mover rights on Windows for key issues.
- Requirement to use some open format document standard. This would allow for alternate companies to access documents, much as we do now with bitmaps, movies, data bases, web pages, etc.
Unless we do something like this, the whole ecconomy can be held hostage by a single company.The same could be said of any of the OSes in the first five years of existance. Windows, that lame dog of an OS from 1984, did not mushroom until after vers 3.x, and Linux did not go anywhere until about 5 years or so later.
I know OS/2 has some bad things about it, but in the main, its quite usable. When you run allocmem it gets rid of the unnecessary dll files that load (specifically, pages them out to swapper.dat). This gives you heaps of core for the apps.
Interesting note: {flamebait}Windows NT derives from v 1.3 of OS/2, and its still there in W2K. I mean, the IBM OS/2 1.3 cmd.exe and rexx runs natively under it. [I've done this, based on a thing in Technet.]{/flamebait}:)
The real reason for wanting an OS/2 laptop is that we can force open standards and a universal installer. For everyone.
But an OS/2 one would be nice :)
The reindeer do not fly. If the flight is coming from the reindeer, then the sleigh would hang. Since it does not hang, the reindeer are normal reindeer, and the whole is set on an invisible road.
Time zones and presumably, the large number of households do not matter. Santa is shown as being back at home on christmas morning. The explination is that santa, like OS/2, is a pre-emptive multitasker.
Once you understand this, then the whole article becomes some sort of skill at showing how the author can do calculations. It's not really good science, since it does not open the consideration that the propositions of the calculations are wrong.
os2fan believes that santa is real (for all that's worth)
In practice, you use alternating columns of 10 and 12, eg 2001 = 1681 = (((1*10+6)*12+8*10)+1). The number is written in any form that makes the parity of the column clear. This means either groups of 2, 4 or 6 digits, or with dots and commas.
The notation I use is V and E for the 10 and 11 digits, and commas and dots to separate different orders of numbers to make them more readable, eg 2**64 = 3.69.01.57.31.E9.50.07.90.16. This can be written as 369.0157.31E9.5007.9016, or as 369,0157.31E9,5007.9016.
The radix is a colon, eg pi=3:16E8.E3....
The number is read using V0 as TEENty, and E0 as eLEFTy, as tenty and eleventy are confusable with twenty and seventy. A hundred is 1.00, a thousand is 1.0000.
Fractions are always read in pairs, not singly, thus pi = three point sixteen, eleftyeight, eleftythree, ...
Zeros in the number are dealt with as follows.
You can see from even this smattering that alternating bases are radically different to the simple bases like 10 and 12.
As an aside, one of the multiplication rules and a certian Linux-friendly penguin share the same name.
Since the year has 12 months, and no month has more than 31 days, you can store a date into a number in the form N = Y*372+M*31+D-32-C, where C is a constant. You can store 89 years into a single signed number, and if you fiddle with negatives, you can get 178 yr, 2 month into a 16-bit number.
If we put packed_date=Y*365+M*31+D-32-C, then we can get Y=(packed_date+C)mod 365, etc. This scheme works because each date has a unique number, although not all numbers correspond to days.
Another trick for the calendar is to start the year on 1 Mar, making the leap year the last day of the year. This makes the length of the month equal to 31-((M-1)mod 5 mod 2).
Also, any proposal that puts days outside the month, eg New year's day not in any month, etc are going to run afoul of the programs that expect days to be in months.
Base 120 beats them all hands down. Not only do all of the numbers 1/(2**a*3**b*5**c) terminate, but 7, 11, 17 have refreshingly short periods. eg 1/7 is 0:17.17.17...,
If you want something smaller, try, eg 18 [where 7*7*7=111, and so 7 divides 1/7, and the same is true for 5 dividing 1/5].
Those of you who think it's a hard base to calculate in have never tried. It's actually easier to work in than base 60.
Historically, it has been used by the Germanic nations, the conversion taking place around 900 onwards. Check out Wright's `Old Norse', where the numbers are given, like "200=one hundred and eighty".
One thing that is easily forgotten is that cardinal time systems are backwards-looking, where ordinals are forward-looking.
In a cardinal system, the elapsed time is given, eg 2000.0000 years. But the point this starts is at the very start of the era, and any point in that 2xxx.xxxx refers to 2000.0000.
In an ordinal system, the elements are named on the part completed one, and so the ends of the era are named, that is, a future point is named. So when we refer to `the third millenium', we refer to `3000', and all dates from the previous millenium point to this, ie '2001-3000'.
The issue here is not so much that `the third millenium starts on 1 jan', but that `if we use a cardinal system, we are harking for the dark past, not the deep future'.
My view on the millenium start is that centuries and higher should carry ordinal names (20 th c, 21th C, 3rd mill), whereas decades and lesser carry ordinal names (eg 1970s, 1963). Moreover, the centuries and millenia align with the decade-starts (so that 21th c runs from 2000.000 to 2099.999).
As to calendar reform, the system is not all that difficult to deal with, and the added variety of the movable feasts and so forth adds a measure of unpredictibility to the calendar. Over here, the late Easter of 2000 ran straight into Anzac day, so we had a five day weekend :) :) :)
There is considerably more to the changing of the calendar than the measurements, since people more often observe the passage of similar numbers in the calendar, than in the linear system. (You do not observe passages of 1 mile or 100 miles after anywhere as often as 1 year or 1 month after).
- Calendars are deeply tied to the religion, an attack on the calendar is seen as an attack on the religion, as the French discovered in 1790
- A fixed week calendar is grossly unfair especially if your birthday falls on a tuesday, and someone else's falls on a sunday each year.
- Investment in the current calendar is high so it is not likely to succeed, because of the associated costs.
None the same, a transitional calendar can be made to work, as long as the two are seen to co-operate. The calendar can expanded in operation, or left as a passing fad. We use base 16 in computers in much the same wayThe planetary names of the week do follow a logical order, though the logic is more obscure today. The seven day week can be made into 10 days by adding days as follows: Monday, Uranus, Tuesday, Wendneday, Neptune, Thursday, Friday, Pluto, Saturday. Unravel these using *7 mod 10, and you will see the Greek order of planets, with the three modern ones at the end.
Clock reform is more closely allied to the weights and measurements. The measurement system works as well under the Christian or Moslem calendars, but fail under a decimal system, because the day is a fixed measure, and measurement systems rely on the second, not the month.
The native metric system is to divide the day into 40 kc , and then into 1000 c, and so forth. Then, the km/kc = m/c. Metric + Decimal Days = unusable, as the km/h is too slow, and the m/s is way too fast.
If you intend to use a hundred-based system (day = 10 hrs of 100 min of 100 sec), the measurement system should be 100-based (eg mile of 100 ch of 100 ft of 100 lines).
Of course, once you invest a large amount of time and effort into databases and scripts that rely on the quirks of some tool (either commercial or open source), you are not keen on tossing the whole lot out for new and better tools. Especially if the new and better tools do not fit into the hole the old one did.
Many of us just like to unwrap the present and play with it, not spend the half-afternoon making the toy and then seeing if it fits our needs. Even if the compiled binary were 85% effective, it will give us an idea of what it does. {sarcasm}The largest software manufacturer in the world makes 85% effective software, and you feed in the other 85%!{/sarcasm}
It would have been interesting to see what the increased productivity was: more lines of code?
It's scary to think that these pesky [ethnics] have been playing around with MS code, and now Credit card data base.
If it gets too bad, we can always go back to the old paper money, and controlled transfers of money. But then the banks would have to employ tellers :(
I mean, it's all well and fine to have rated pages. We need to take the next step, which should be like this:
Imagine a browser ratings engine that works off a standard list. You download a rating-profile from your site. It says `allow x, deny y'. Were the ratings-table a standard format, then you can download for any browser, and any society, a rating table. For example, you could use Netscape and a AntiScientology page, or a `no evolution' page. You should be able to load multiple issues, eg no-evolution + no-commie-stufff.
Maybe we could edit our own as well ...
The thing is, if it's there and not working, it might as well not work. If we want to make it work, then we need to push the UI somewhat.
If you look at some hotel guide, you see all of these silly little icons, like `food', `showers' etc. You can work out what facilities are available from these.
Now imagine something like a search engine, picking up these ratings icons, and filtering them according to requests. You want no {x}, we'll screen out the {x} or grey them out or whatever your preference is.
Site rating would be done at the first at peer level. There would be some rights of appeal, to a certian number of authorities. There would be about three or four levels of grievence appeals each costing additional money.
Sites can decline to participate, and you can decline to use the ratingware on your system. But in both cases, it's like going outside. You don't know what sort of strangers you will meet... Sites should also be able to reject an applied rating (as in opt out), even to the extent of posting a `don't disturb' sign on the door.
Say I create a site "occulthelp.org". It may get rated as some setting, eg "occult" by things that look for sting. On the other hand, the site might be a genuine resource for families that are recovering from excessive occultdom in one of the children.
If the site were subjected to peer review, the rating might reflect the true help-recover-from nature of that site, rather than a `this is more'.
You could have three or four or so chanels of censorship, by different groups. You as a user might then select the ratings engine as you select the search engine.
I mean, if we want anything better than big business, grep or whatever looking after our ratings, we have to do something ourselves on it. Hence this suggestion ... :)
It sounds pretty optional, but it is a lot better than name-guessing, grepping and whatever. I mean, if you as a page opt out, you can not be seen by those who have ratingware running. It's your call, not someone else's
Peope are looking for some sort of content control... Either
- we're going to leave it to the big end of town to decide what we watch, or
- netizens are going to police the rating scale, or
- we're going to run a zillion standards that people are going to make uniform, or
- [Heaven forbid] they might kill off the net
...
In any case, if we don't make a move and get some sort of ratings systems, (any ratings systems), we are gong to live of grep's rating.It's something like they do for the movies.
An interface should give the user some indication on what to do, and somewhere to do it. The basic function of what to do could be explained in the manuals, like mouse clicks, etc. But the rest of it should be visible.
For example, the CLI emulates a teletype, with a few standard hacks. The CLI brings to the TTY, the notion of redirection of output, a programming interface, and pipelineing. Once mastered, the CLI can prove to be a powerful workhorse.
In a windowing system, you can run multiple CLI sessions. You would be very upset if commands made in one session affected another. This is called leaking. Yet, different command lines could be logged to the same log file, what is going on would be clear if different names are used: consider the IRC sessions, for example.
What makes GUIs scary is that it is all too easy to hide things, and move them from place to place. The interface varies from user to user, and from vendor to vendor, and from version to version.
A small UI, like the Win3x shell, makes a dandy thing if the main activities is to launch intensive applications, like games, or to load different users, or to install hardware and software to the OS. The shell is small, and fits nicely into ram, and is more userfriendly than a command line. A fixed shell makes it easier to describe in manuals, for one-time tasks like installation, or rare activities like fs maintainance.
Once into an activity, you can then load the shell of your choice (on a user by user basis), and your customisations for that user. This can be a full set of widgets and hacks designed to allow programs to integrate.
What happens now is that each new version of Windows seems to say `lets make Windows look like OS/2 v 1 [so we got Program Manager and File Manager]. Let's make it look like v 2.0, so we get Explorer. Let's make it look like Netscape, so we get the Win98 explorer thing. And let's make it look like the Adobe interface, so we get this Win.NET thing.
So if you want to do any nice things, you have to learn the weird languages that make the different shells tick, or just leave them alone.
And so you see why, despite the power of the shells, most people use them in much the same way as the dodgy little menus that came with the DOS machines in the late 80's. Use the shell to start programs. All the rest of the shell is so much lard.
We are seeing the leading edge of what is now large scale computer snooping, eg Carnivore.
Eventually, a lot of crunching external data may become legit, like SETI@home is. It's not that far off that we'll start seeing `commercial in confidence' snooping.
Also, you don't need to nail the person on the collected data, all you need to do is alert that this person is worth watching, and nail on some legal-to-collect thing.
And being companies, there is little around the commercial in confidence stuff, and even less if there is a one-way ID system. [eg file `os2fan' contains no references to file `votegeek', but file `votegeek' contains commentry on `os2fan'.
Scary, hey.
But that does not stop others ....
It seems whoever moves the learning makes the rules.
Think of a computer connected to a tty machine. You have a program looking after the console. It reads input, it runs output.
- You can build in commands to it.
- You can endower a simple or complex macro language to it.
- You can allow it to store output to other devices, or files.
- You can allow it to gobble up bits of the output as input.
Sounds like a command shell to me, like 4os2, command.com or csh or bash. These differ in what they can do, how good is their macro language, or the built inservices, but they are much the same otherwise.The UI has been around a long time before computers, and guess what? - it works. It even works under a GUI. And, you will find that users would get terribly upset if the pipe or keystrokes from one tty session bled through to another ...
So yes, in the beginning there was a lonely command line. But the command line needed to talk to a something. At first it phoned up other command lines, and talked to them. So was born the telegram, the telex, and the teletype.
May years passed, and they needed to talk to computers. Computers could not talk, but they could type, and who better than a telex to talk to. So we started by telexing the computer.
Telexes were expensive, and most users had to use punched cards. But when the teletype terminals became cheaper, the punched cards disappeared, and so on.
Terminals became more capable, and one could deal with the terminal as a changing page, rather than a tty session. You could use the screen as a window on a page, and from this, via a long and winding road, we get character based terminal sessions, windows, and finally, the ultimate in perfection for the GUI, the workplace shell ...
I really don't know the answer myself, but I'll toss in some thoughts. Consider for example, some sort of pool of money collected from the distribution of media. As you buy the media, you acquire a certian licence to hold a volume of MP3s. Artists would be paid some sort of royalty based on much the same way as radio royalties, with some additional basis for the `underground' stations.
I know it may hurt, but I'm afraid that if the copy process can not be trusted, then that's the way things are going to head.