What would the advantages of putting an ARM processor over a Crusoe in a PDA? If the Crusoe can morph any instruction set, wouldn't it be able to run ARM instructions? How would a company benefit from using ARM chips in their PDAs, other than a slight decrease in power consumption?
Well, for a start, the SA-1100 uses less power at 200MHz (about 200mW) than the Crusoe at 400MHz (about 1W). Even given that power consumption tends to go as voltage squared, and that still means that the SA core gives you more MIPS/Watt. Secondly, unless the Crusoe RISC core matches well against the SA core in as much as having virtually equivalent instructions, emulating a RISC instruction set with a different RISC instruction set can lead to performance hang ups. For example, the ARM chips tend to have both conditional instructions and barrel shifters on each instruction, so while they are single issue they can end up doing several things for instruction. (Rusty ARM code warning alert!!)
e.g. ADDS r1,r1,r2 LSR#2; MOVNE r0, r1 LSL#4
Add r1 and r2DIV4 , place in r1 and set flags on result of sum. If negative flags set, move r1 shifted left 4 into r0.
As you can see, if for each of these instructions, you end up firing off two or three Crusoe RISC core instructions, performance can be degraded to a level at around a third of the base clock speed. Similar problems do not occur to the same extent with CISC, as you have more scope to interpret and pass on RISC instructions from CISC ones (as I believe AMD do in their intel compatible chips).
The other thing that springs to mind is that if Linux is your target OS, then either Crusoe running Linux under x86 emulation or a StrongARM running ARMLinux will fit your requirements well, so if battery life is your ultimate goal, then the StrongARM hardware may fit your requirements better.
Just to prove that you really can spin statistics anyway you choose, lets take a look through some of those questions and answers.
Question 1: STATEMENT: Increased government regulation and litigation of the Technology industry will lead to consumers paying higher prices.
Loaded question number one. Litigation => lawyers. Higher costs are inevitable in this scenario, and not surprisingly, over half the respondents thought this was the case. Had they bothered to separate regulation from litigation it's not so clear what the answer would be.
Question 2: Do you think increased government regulation, including the regulation of software design, will have a negative or positive effect on the high tech industry and companies, like Microsoft, to innovate and bring new products to consumers?
This is another interesting slant. Regulation of software design? Wow. Just load those questions with negative impact - the idea of regulating software 'design' would have the whole of Slashdot up in arms faster than a DVD case. Funnily enough, over half the respondants felt this was a bad thing too.
Question 3: The U.S. Justice Department is currently suing Microsoft. How closely are you following the trial - very closely, somewhat closely, or not closely at all?
Now this is an interesting question. And the answer is very revealing - only 5% of the respondants were paying close attention to the trial. So this survey, even given it's headline claim of 63% Americans thought that breaking up MS was a bad thing, actually managed to pick a group of people who were split between having occassionally heard of the MS vs DOJ case and those who haven't followed the case at all.
I could go on. Most of the questions are too loaded to be worth discussing. If you are planning on spending money on a real survey, just remember that you must not lead the interviewee by loading up your questions one way. I would also suggest asking questions in random order. Once you find someone leaning one way at the start of a questionnaire, people tend to follow that path. So, had the questionnaire started with:
Question: If there were more alternatives to the Microsoft Windows platform available to you at lower cost, would you be interested?
... and...
Question:If the MS vs DOJ trial results in Microsoft being broken up and you end up having a wider choice of quality software for your computer as a result of the increased competition in the software market, would you be interested?
One of Mike's key theories seems to be that there is no sense of feedback between the end-user and the coder in Open Source development. I would refute this fairly strongly - often the people I know using Open Source tools have fairly widespread experience of other User Interfaces, ranging often both across multiple platforms and going way, way back to before the days when the Graphical User Interface first raised it's head above the primordial digital soup. Now this experience does not make any of these people a UI expert, nor does it necessarily mean that the programs they write have well designed User Interfaces. It does however give us the possibility of recognising good UI design when we get to experience it, and also the possibility of influencing the design of the User Interface in later releases.
In these days of expanding user-base for Linux, and the push to provide a more newbie-friendly environment to work in (which, by the way, I totally support), good User Interface design is getting to be much more important. There are various resources appearing, from the Gnome UI Improvement project and its mailing list, along with the work that the KDE people are putting together with KDE 2.0, which are testament to the need to try and learn from the many graphical interfaces out there and to innovate as well. Having a well designed UI need not reduce the speed at which the experienced user uses their machine, while allowing the novice some hope of making progress.
Innovation is often overlooked in designing a new UI. As soon as you stray from, say, the way MS Windows does something, people jump up and down worrying that new users will be confused by a different method. I'd disagree - just because it has been done that way before is not, in itself, reason to continue doing it. A good UI *must* be intuitive and logical at some level - simply copying the existing behaviour of other window managers will not end up with a coherent project. At the moment, the graphical user interface is a mess of conflicting ideologies. We all have experienced the frustration of 'Drag-and-drop' when it isn't a universal quality - for example in Windows, you can (sometimes...!) drag a file into a program to load it, but you can't drag that file out to save it or pass it to another application to work on it in a different way. And I don't mean using the clip board either, although that may be the route that would be used to effect such a transfer, it shouldn't be obvious to the user that that is how it happened.
Taking the best paradigms for working with a graphical user interface and making it all stick together in a cohesive fashion is a task of iteration, experience and reiteration between the end-user and the coder. Since in the Open Source world the user may also wear the coders hat, this should be the ideal environment in which to create and refine the most useable graphical interface on any platform, as long as we keep our sights on some central game plan of Useability and not merely on creating a feature-rich tick list of things our programs can do.
However, I'm concerned that some of these hardware companies aren't releasing the source of the drivers they're providing. I'm not sure why they want to keep their source closed, because I personally can't see any disadvantage in opening the source for a hardware driver - after all, it's in selling the piece of kit that they make their money, not selling the driver, surely.
I wonder whether Aureal have a problem with the way A3D is supported on the older cards. As far as I know, the Vortex 1 (AU8820) supports A3D 1.0 in hardware, but relies on the driver to provide A3D 2.0 if it is requested. The Vortex 2 (AU8830) does A3D 2.0 in hardware, but now that there is A3D 3.0 in the works, I wonder whether we will see a software implementation of A3D 3.0 for the AU8830 chipset. The cost with a semi-hardware solution is, of course, CPU time.
What this boils down to is that to avoid putting any A3D code (and in effect Intellectual Property) into the drivers, there will be a need to provide separate drivers for each chipset. At the moment, I'm happy that my soundcard (which came bundled with my machine) is having drivers developed by the manufacturer for Linux. And I hope that sooner or later, we will see source code releases for each chipset to take advantage of the features available from the hardware, albeit without the latest version of A3D software patched over the top.
The game engine is a direct 3d game engine. They are going to have a hard time. Since they are probably going to port it to openGL they can probably also get an openGL version for windows.
Myth II, ported by Loki Software from the original by Bungie Software, ran under Direct3D in the Windows version (I have no idea what it ran on the Mac version), so this sort of problem has already been solved at some level. I note that Lithtech already have ports to platforms other than Windows anyway (Apple Mac and Amiga) due to a deal made with Hyperion Software in April 1999, so there must be a reasonably clean level of separation between the graphic rendering side and the underlying engine.
What does CorelDRAW provide that GIMP doesn't (or couldn't)?
CorelDraw 9 is actually a small suite of packages, including CorelDraw, Corel Photo-Paint, a font navigator, a texture explorer, a bitmap-to-vector tracing package and various image distortion tools. So, to answer your question, the functionality provided by CorelDraw 9 that the GIMP doesn't do is vector-based artwork, rather than pixmap. This is still an area of the Linux application base that is not fully up to speed yet - there are various applications which do vector-art/vector-design on Linux, such as Dia, Sketch, KIllustrator, Xfig (ancient but still useful) and it's successor GTKFig, GYVE and Impress but many of these are as yet incomplete or have fallen by the wayside. That's not to say that CorelDraw 9 is necessarily the best vector art package out there - I'd like to see the latest Adobe Illustrator on Linux too - but it is a welcome filling-out of the application base.
There are several things in the Windows package which it will be very interesting to see what Corel do with regards to porting them, or if they are simply ommitted. For example, the MS Visual Basic for Applications scripting language used for automation of CorelDraw 9 - drop or replace? - and the Digimarc Digital Watermarking software, something I'm currently unaware of anything like this on the Linux platform. Plus the usual glut of a thousand TrueType and Type1 fonts you get with any vector or DTP package these days.
Whether Corel Photo-paint 9 holds a candle to the GIMP (I don't honestly know, since I haven't used Photopaint since v5) is vaguely irrelevent, since it is the vector art package in this lot that will probably be of most interest to most people.
There doesn't seem to be too many signs of impending technology recession where I'm sitting. Given that IBMs net income for Q4 in 2000 were up 28% over the previous year, that doesn't exactly sit with the tech doom and gloom. Even if you take the whole year 2000, IBM posted a 16% increase over 1999 despite all the muttering that 2000 was a slow year.
Sounds like he's suggesting a moderated on-line science journal with two types of users: those with accounts (graduate students of various fields who have the permissions to moderate within their field) and those without accounts who essentially read-only.
Having habitually waded through the thirty or forty new electronic abstracts available through one of the astronomy preprint servers each morning during my Astronomy PhD, having some sort of moderated, sorted, categorized, up-to-the-minute Science Journal sounds like heaven. Of course, I've now left my astronomy roots to work at IBM Canada, but having spent four years of my life submerged in Astronomy still leaves a significant urge to continue watching the science evolve and mutate.
And herein lies the problem - what one person in a scientific discipline considers to be a good paper may be considered a complete dud by another. A classic example from my subject from a couple of years back is the Hubble Constant. For the uninitiated, this is the constant that represents the ratio of the velocity of a galaxy rushing away from us to the distance to that galaxy. In an expanding universe at one point in time, it's constant at large enough scales... we think... if the galaxy isn't gravitationally coupled to the viewer, or if it isn't involved in some super-cluster super-flow... But anyway, I digress. The value of the Hubble constant, very much a 'Holy Grail' of observational astronomy, is difficult to measure accurately. Various methods give different results - the lab I worked at used a method based on the Sunyaev-Zel'dovich effect and had a best fit answer of 42 km s-1 Mpc-1. Various other groups, working from Cepheid variable stars, had values up in the 75 - 100 km s-1 Mpc-1. So our group naturally tended to look more favourably on papers with lower values (say less than 60) and other groups would prefer high values. So any form of moderation is extremely sensitive to who is moderating it - there are egos in science just like any other field!
At the end of the day, this therefore requires some sort of anti-bias measure. Having lots of people moderate the papers is therefore a good thing, but that requires significant man-power. Something similar to the Slashdot scheme might work, but the number of people who are in a position to make intelligent moderation on a highly technical paper is much smaller than those who might moderate more general material, and that small group may have better things to do than read 40 papers a day!
Even so, some sort of moderated online science journal scheme would be interesting, despite the problems... It might also encourage an increase in the lucidity of scientific papers in order to get moderated up - it might even generate a race to see who could notch up the highest Slashdot^H^H^H^H^H^Hcientific Karma:-)
If I'm reading this correctly, the patent claims a handwriting recognition scheme that uses (graphically well-separated) characters that can be created with unbroken strokes. Said characters are allowed to have no spatial relation to each other.
This patent strikes me as being similar to the approach adopted by banks for reading cheque and account numbers off chequebooks (at least in the UK). The numbers 0-9 are distorted and reworked slightly so that automated readers have a much lower chance of making a mistake. This patent does not seem to me to go much beyond common sense either. Even trivial attempts at a handwriting recognition system using pixelated-probability maps to identify characters have huge difficulties in separating letters like X and Y, or O and Q, and for successful handwriting recognition you have to break these degeneracies at some point. The simplest method is simply to make sure that a Y can't look like an X and that relies on the user to make that step. If this is valid for a patent, maybe there are some more simple ideas I should start looking to patent myself!:-)
It will be interesting to see whether the court upholds this patent, or whether it is considered to be too broad to be patentable. I note that at this stage all that has been agreed is that the patent can be used to bring a case, so there is no guarantee for Xerox that this patent will be valid once the court has finished with it.
If I understand preferential stock prices at all, there will almost certainly be a time period during which you can't turn round and sell your newly acquired shares. Without knowing what restrictions apply for MS employees, I'd guess that they have to wait between 6 months and a year before being able to sell them. Who knows where MSFT will be in a year? Analysts are predicting a share price near US$200 for the end of 2000, but I'd say that the current spate of MS bad fortune may be causing some hasty revisions of projections in many funds. But historically, MS has always managed to support its share price well year-on-year. After all, from an employee perspective, there is a certain element of golden handcufts with stock options - you always want them to mature before you leave, but if the share price drops sharply, things don't look so rosy any more. In many ways, a long term drop in the share price of MSFT would be a very strong indicator that the MS "World Domination Plan" (TM) is losing its grip.
Regarding anti-aliased fonts: There's actually a simple way to do that, but it eats RAM like a wumpus. The simple way is not to use fonts at all, but instead to create pixmaps (e.g. XPM format) that have anti-aliased text. This would only be useful on a system with lots(lots) of RAM and then only useful for largely static content like menus, dialogs etc.
I've used a system (Acorn RISC OS) which makes extensive use of fully-anti-aliased fonts (including sub-pixel alignment both vertically and horizontally) and its solution was both elegant and not a memory hog. Starting with vector-based fonts (I'll call them outline fonts) which additionally had hint lines so that thin parts of the letters were not lost at small font sizes, it built pixmaps on the fly and cached them in a pre-allocated font cache. As the text was written to the screen, pre-pixmapped fonts were pulled from the cache and plotted direct: where pixmaps had not been calculated, a block (32?) of characters were rendered into the cache from the relevent font and then plotted. Fonts above a user-defined point-size limit were always calculated and plotted direct from the outline font rather than being cached as pixmaps, and vertical and horizontal sub-pixel alignment could be toggled on and off at will. When plotting the anti-aliased fonts, the pixmap was alpha-blended with any background texture present to avoid strange artefacts. For most of the time, 512K was complete overkill for the font-cache size, and since most documents contain maybe 96 characters in two fonts in two font sizes, the pixmaps don't take up as much space as you might expect. If you change the font in the document, the new font will end up in the cache, and the old one may be discarded if the cache is full, or left in place if the space allows.
Anti-aliased fonts make a huge difference to font readability when done properly, and make a significant impact when sub-pixel alignment is enabled for viewing whole pages of text. No more lines of text separated by one extra line of white pixels, or strange staccato problems when faced with a row of......... I often had friends comment that the text looked like it had been scanned on a high quality scanner from some postscript output, rather than rendered on the fly, and scrolling speeds on a 200MHz machine rarely, if ever, hiccupped except on documents with vast collections of fonts and font sizes.
Corel Linux has been out for some time and I'm still bothered that a company like Corel wants to take their programmers off other projects and reinvent a *better* Debian distro.
Corel has had to work hard on a Linux distro because it has had to build one for its ARM-Linux-based Netwinder machines. All that time and experience in building a non-Intel Linux platform distro is reasonably reinvested in getting an Intel-distro put together to expand their platform presence and provide themselves with a well-understood distro on which to test their software. After all, even in these enlightened platform-independent days (I wish...) Linux on x86 is still the most widespread, and any commercial binary-only releases of software (such as CorelDraw 9) will have to appear on the x86 Linux to see reasonable sales. I strongly hope that they will compile up versions for other Linux platforms as well, but I won't hold my breath for those to appear in the local store - I'd expect those to be dispatched from Corel itself on order.
To be honest, I don't think it is just the FPS shooters which are attractive to the Linux crowd. In fact, the history of games on Unix in general seems to suggest that almost anything goes, from the phenomenal longevity of Xpilot with its fast multiplayer action, to the Rogue-like collection (most notably Angband and Nethack, along with Omega and others) and of course the many MUDs. In my opinion, the balance seems to be more towrds thinking, strategy games than anything else.
Now, the reasons for this trend so far may well be that vt terminals are fairly universal, and that the games in the past have been developed for universality. Looking at the current state of development in the Linux game web pages, such as LinuxGames and HappyPenguin, I see various hopeful signs that 3D is really picking up (Parsec, Xracer) and the arrival of Xfree86 4.0 will further enhance developments in this area.
One of the weaknesses of the Linux game development is that the big companies are still too wary of getting into producing Linux games. Loki is doing a great job in getting companies to give the go-ahead for porting games. For the foreseeable future I suspect that Loki's expertise in the Linux field will continue to make this an easier route for games to appear on the Linux platform. It will be a while before Linux expertise is widely grounded inside commercial game companies and Linux versions are produced in-house, although obviously id software has no problems in this area!
AMD do have a 1GHz Athlon on the market - it's in the Kryotech SuperG machine, retailing at $2200 I think and available now.
Since Kryotech stuck with AMD through the leaner periods until now, it's nice to see their cooling technology get a worthy processor.
Of course, the general media haven't caught up to this, which is odd given that there was a howling crowd of posters in one of the talkbacks to an article on ZDNet looking at the race to 1GHz all wondering why the Kryotech machines aren't being more widely publicized. No wonder I keep bumping into conspiracy theories about Intel 'warning off' OEMs and PC builders...:-)
With so many sites now offering dynamic, up-to-the-minute information, merely caching the contents of a page at a particular moment in time is at best only catching glimpses of these pages, and at worst leads to misleading search results when querying the search database. It strikes me that with these sorts of site, where the content of the pages is changing so rapidly, something more objective is needed.
For example, the first clues about the state of flux of a particular page can be obtained by diff'ing the page against the previous copy held in memory. If the page is simply having extra items added in periodically, such as a FAQ, then the diffs will generally show that there is more being added than taken away, and the traditional snapshot method employed by search engines is fine. However, if the page is wildly different in the majority of its content (such as the slashdot main page), there is practically no point in making a copy of the page for indexing purposes. A much better solution is attempt to build a keyword database automatically for this page by lexical analysis of the text - even a '100-most-common-words' list (with 'the', 'and', 'is' etc filtered out) would be an improvement on the current situation. As repeated visits build up, this keyword list will refine itself and actually provide some reasonable pointers to the material likely to be found at the site.
With any large database, particularly when you get to the stage where GB's of information are consider small fry, the need for efficient data mining and generation of useful indices becomes increasingly important. Database technologies are looking forward to a time when there will be a need for Petabyte and Exabyte storage and retrieval, and effective distillation of a web page's information, rather than a word-for-word verbatim cache will be the only answer.
Jon Katz raises all sorts of ethical worries surrounding the Human Genome project, but I feel he misses the point.
The Human Genome Project sets out to *map* the Human Genome - it does not set out to fully understand it. What it does give is a wealth of reference material to help researchers fighting disease and disability try and identify common traits in their subjects.
The Human Genome Project is *not* gene-therapy. It is *not* the new Frankenstein. It is not even being able to choose the appearance of your offspring. Too often the media ends up mistaking information about something as the means to actually do something - a bit like the difference between knowing that sufficient plutonium in the right place can go critical and actually trying to make an effective nuclear weapon. Witness the attempts made by governments all over the world to get a nuclear program going, and the number of successes. The theory is well understood - the practice is more difficult.
Will the Human Genome project change our lives? Possibly. But only as the result of building on the database of information it gives us. The opportunity to allow the development of therapies to improve the quality of life for many people should not be missed. I agree that there are many applications of this information that are ethically-questionable, or morally repugnant, but the Human Genome Project is not the application - it is merely the reference.
Maybe there should be a wider debate on the ethics of genetics, and genetic applications. It would be welcomed to see an informed and intelligent discussion, but with the current media's fixation with hyperbolae and sensation I fear that the important issues will once again get buried beneath an avalanche of "Frankenstein Foods" and similar headlines. Maybe that is an excessively cynical view but I repeatedly see important issues obscured by headline-grabbing stories running on people's fears. In this technological age, the public understanding of science is too often blurred and confused by scare stories in the general press - oh for a day when people are presented with unbiased information with which to make up their own minds.
Slashdot Mirror
What would the advantages of putting an ARM processor over a Crusoe in a PDA? If the Crusoe can morph any instruction set, wouldn't it be able to run ARM instructions? How would a company benefit from using ARM chips in their PDAs, other than a slight decrease in power consumption?
Well, for a start, the SA-1100 uses less power at 200MHz (about 200mW) than the Crusoe at 400MHz (about 1W). Even given that power consumption tends to go as voltage squared, and that still means that the SA core gives you more MIPS/Watt. Secondly, unless the Crusoe RISC core matches well against the SA core in as much as having virtually equivalent instructions, emulating a RISC instruction set with a different RISC instruction set can lead to performance hang ups. For example, the ARM chips tend to have both conditional instructions and barrel shifters on each instruction, so while they are single issue they can end up doing several things for instruction. (Rusty ARM code warning alert!!)
e.g. ADDS r1,r1,r2 LSR#2; MOVNE r0, r1 LSL#4
Add r1 and r2DIV4 , place in r1 and set flags on result of sum. If negative flags set, move r1 shifted left 4 into r0.
As you can see, if for each of these instructions, you end up firing off two or three Crusoe RISC core instructions, performance can be degraded to a level at around a third of the base clock speed. Similar problems do not occur to the same extent with CISC, as you have more scope to interpret and pass on RISC instructions from CISC ones (as I believe AMD do in their intel compatible chips).
The other thing that springs to mind is that if Linux is your target OS, then either Crusoe running Linux under x86 emulation or a StrongARM running ARMLinux will fit your requirements well, so if battery life is your ultimate goal, then the StrongARM hardware may fit your requirements better.
Cheers,
Toby Haynes
Question 1: STATEMENT: Increased government regulation and litigation of the Technology industry will lead to consumers paying higher prices.
Loaded question number one. Litigation => lawyers. Higher costs are inevitable in this scenario, and not surprisingly, over half the respondents thought this was the case. Had they bothered to separate regulation from litigation it's not so clear what the answer would be.
Question 2: Do you think increased government regulation, including the regulation of software design, will have a negative or positive effect on the high tech industry and companies, like Microsoft, to innovate and bring new products to consumers?
This is another interesting slant. Regulation of software design? Wow. Just load those questions with negative impact - the idea of regulating software 'design' would have the whole of Slashdot up in arms faster than a DVD case. Funnily enough, over half the respondants felt this was a bad thing too.
Question 3: The U.S. Justice Department is currently suing Microsoft. How closely are you following the trial - very closely, somewhat closely, or not closely at all?
Now this is an interesting question. And the answer is very revealing - only 5% of the respondants were paying close attention to the trial. So this survey, even given it's headline claim of 63% Americans thought that breaking up MS was a bad thing, actually managed to pick a group of people who were split between having occassionally heard of the MS vs DOJ case and those who haven't followed the case at all.
I could go on. Most of the questions are too loaded to be worth discussing. If you are planning on spending money on a real survey, just remember that you must not lead the interviewee by loading up your questions one way. I would also suggest asking questions in random order. Once you find someone leaning one way at the start of a questionnaire, people tend to follow that path. So, had the questionnaire started with:
Question: If there were more alternatives to the Microsoft Windows platform available to you at lower cost, would you be interested?
... and ...
Question: If the MS vs DOJ trial results in Microsoft being broken up and you end up having a wider choice of quality software for your computer as a result of the increased competition in the software market, would you be interested?
... the results might just have been different :-)
Cheers,
Toby Haynes
One of Mike's key theories seems to be that there is no sense of feedback between the end-user and the coder in Open Source development. I would refute this fairly strongly - often the people I know using Open Source tools have fairly widespread experience of other User Interfaces, ranging often both across multiple platforms and going way, way back to before the days when the Graphical User Interface first raised it's head above the primordial digital soup. Now this experience does not make any of these people a UI expert, nor does it necessarily mean that the programs they write have well designed User Interfaces. It does however give us the possibility of recognising good UI design when we get to experience it, and also the possibility of influencing the design of the User Interface in later releases.
In these days of expanding user-base for Linux, and the push to provide a more newbie-friendly environment to work in (which, by the way, I totally support), good User Interface design is getting to be much more important. There are various resources appearing, from the Gnome UI Improvement project and its mailing list, along with the work that the KDE people are putting together with KDE 2.0, which are testament to the need to try and learn from the many graphical interfaces out there and to innovate as well. Having a well designed UI need not reduce the speed at which the experienced user uses their machine, while allowing the novice some hope of making progress.
Innovation is often overlooked in designing a new UI. As soon as you stray from, say, the way MS Windows does something, people jump up and down worrying that new users will be confused by a different method. I'd disagree - just because it has been done that way before is not, in itself, reason to continue doing it. A good UI *must* be intuitive and logical at some level - simply copying the existing behaviour of other window managers will not end up with a coherent project. At the moment, the graphical user interface is a mess of conflicting ideologies. We all have experienced the frustration of 'Drag-and-drop' when it isn't a universal quality - for example in Windows, you can (sometimes...!) drag a file into a program to load it, but you can't drag that file out to save it or pass it to another application to work on it in a different way. And I don't mean using the clip board either, although that may be the route that would be used to effect such a transfer, it shouldn't be obvious to the user that that is how it happened.
Taking the best paradigms for working with a graphical user interface and making it all stick together in a cohesive fashion is a task of iteration, experience and reiteration between the end-user and the coder. Since in the Open Source world the user may also wear the coders hat, this should be the ideal environment in which to create and refine the most useable graphical interface on any platform, as long as we keep our sights on some central game plan of Useability and not merely on creating a feature-rich tick list of things our programs can do.
Cheers,
Toby Haynes
However, I'm concerned that some of these hardware companies aren't releasing the source of the drivers they're providing. I'm not sure why they want to keep their source closed, because I personally can't see any disadvantage in opening the source for a hardware driver - after all, it's in selling the piece of kit that they make their money, not selling the driver, surely.
I wonder whether Aureal have a problem with the way A3D is supported on the older cards. As far as I know, the Vortex 1 (AU8820) supports A3D 1.0 in hardware, but relies on the driver to provide A3D 2.0 if it is requested. The Vortex 2 (AU8830) does A3D 2.0 in hardware, but now that there is A3D 3.0 in the works, I wonder whether we will see a software implementation of A3D 3.0 for the AU8830 chipset. The cost with a semi-hardware solution is, of course, CPU time.
What this boils down to is that to avoid putting any A3D code (and in effect Intellectual Property) into the drivers, there will be a need to provide separate drivers for each chipset. At the moment, I'm happy that my soundcard (which came bundled with my machine) is having drivers developed by the manufacturer for Linux. And I hope that sooner or later, we will see source code releases for each chipset to take advantage of the features available from the hardware, albeit without the latest version of A3D software patched over the top.
Cheers,
Toby Haynes
The game engine is a direct 3d game engine. They are going to have a hard time. Since they are probably going to port it to openGL they can probably also get an openGL version for windows.
Myth II, ported by Loki Software from the original by Bungie Software, ran under Direct3D in the Windows version (I have no idea what it ran on the Mac version), so this sort of problem has already been solved at some level. I note that Lithtech already have ports to platforms other than Windows anyway (Apple Mac and Amiga) due to a deal made with Hyperion Software in April 1999, so there must be a reasonably clean level of separation between the graphic rendering side and the underlying engine.
Cheers,
Toby Haynes
What does CorelDRAW provide that GIMP doesn't (or couldn't)?
CorelDraw 9 is actually a small suite of packages, including CorelDraw, Corel Photo-Paint, a font navigator, a texture explorer, a bitmap-to-vector tracing package and various image distortion tools. So, to answer your question, the functionality provided by CorelDraw 9 that the GIMP doesn't do is vector-based artwork, rather than pixmap. This is still an area of the Linux application base that is not fully up to speed yet - there are various applications which do vector-art/vector-design on Linux, such as Dia, Sketch, KIllustrator, Xfig (ancient but still useful) and it's successor GTKFig, GYVE and Impress but many of these are as yet incomplete or have fallen by the wayside. That's not to say that CorelDraw 9 is necessarily the best vector art package out there - I'd like to see the latest Adobe Illustrator on Linux too - but it is a welcome filling-out of the application base.
There are several things in the Windows package which it will be very interesting to see what Corel do with regards to porting them, or if they are simply ommitted. For example, the MS Visual Basic for Applications scripting language used for automation of CorelDraw 9 - drop or replace? - and the Digimarc Digital Watermarking software, something I'm currently unaware of anything like this on the Linux platform. Plus the usual glut of a thousand TrueType and Type1 fonts you get with any vector or DTP package these days.
Whether Corel Photo-paint 9 holds a candle to the GIMP (I don't honestly know, since I haven't used Photopaint since v5) is vaguely irrelevent, since it is the vector art package in this lot that will probably be of most interest to most people.
Cheers,
Toby Haynes
There doesn't seem to be too many signs of impending technology recession where I'm sitting. Given that IBMs net income for Q4 in 2000 were up 28% over the previous year, that doesn't exactly sit with the tech doom and gloom. Even if you take the whole year 2000, IBM posted a 16% increase over 1999 despite all the muttering that 2000 was a slow year.
Food for thought.
Toby Haynes
Sounds like he's suggesting a moderated on-line science journal with two types of users: those with accounts (graduate students of various fields who have the permissions to moderate within their field) and those without accounts who essentially read-only.
Having habitually waded through the thirty or forty new electronic abstracts available through one of the astronomy preprint servers each morning during my Astronomy PhD, having some sort of moderated, sorted, categorized, up-to-the-minute Science Journal sounds like heaven. Of course, I've now left my astronomy roots to work at IBM Canada, but having spent four years of my life submerged in Astronomy still leaves a significant urge to continue watching the science evolve and mutate.
And herein lies the problem - what one person in a scientific discipline considers to be a good paper may be considered a complete dud by another. A classic example from my subject from a couple of years back is the Hubble Constant. For the uninitiated, this is the constant that represents the ratio of the velocity of a galaxy rushing away from us to the distance to that galaxy. In an expanding universe at one point in time, it's constant at large enough scales... we think ... if the galaxy isn't gravitationally coupled to the viewer, or if it isn't involved in some super-cluster super-flow... But anyway, I digress. The value of the Hubble constant, very much a 'Holy Grail' of observational astronomy, is difficult to measure accurately. Various methods give different results - the lab I worked at used a method based on the Sunyaev-Zel'dovich effect and had a best fit answer of 42 km s-1 Mpc-1. Various other groups, working from Cepheid variable stars, had values up in the 75 - 100 km s-1 Mpc-1. So our group naturally tended to look more favourably on papers with lower values (say less than 60) and other groups would prefer high values. So any form of moderation is extremely sensitive to who is moderating it - there are egos in science just like any other field!
At the end of the day, this therefore requires some sort of anti-bias measure. Having lots of people moderate the papers is therefore a good thing, but that requires significant man-power. Something similar to the Slashdot scheme might work, but the number of people who are in a position to make intelligent moderation on a highly technical paper is much smaller than those who might moderate more general material, and that small group may have better things to do than read 40 papers a day!
Even so, some sort of moderated online science journal scheme would be interesting, despite the problems ... It might also encourage an increase in the lucidity of scientific papers in order to get moderated up - it might even generate a race to see who could notch up the highest Slashdot^H^H^H^H^H^Hcientific Karma :-)
Cheers,
Toby Haynes
If I'm reading this correctly, the patent claims a handwriting recognition scheme that uses (graphically well-separated) characters that can be created with unbroken strokes. Said characters are allowed to have no spatial relation to each other.
This patent strikes me as being similar to the approach adopted by banks for reading cheque and account numbers off chequebooks (at least in the UK). The numbers 0-9 are distorted and reworked slightly so that automated readers have a much lower chance of making a mistake. This patent does not seem to me to go much beyond common sense either. Even trivial attempts at a handwriting recognition system using pixelated-probability maps to identify characters have huge difficulties in separating letters like X and Y, or O and Q, and for successful handwriting recognition you have to break these degeneracies at some point. The simplest method is simply to make sure that a Y can't look like an X and that relies on the user to make that step. If this is valid for a patent, maybe there are some more simple ideas I should start looking to patent myself! :-)
It will be interesting to see whether the court upholds this patent, or whether it is considered to be too broad to be patentable. I note that at this stage all that has been agreed is that the patent can be used to bring a case, so there is no guarantee for Xerox that this patent will be valid once the court has finished with it.
Cheers,
Toby
If I understand preferential stock prices at all, there will almost certainly be a time period during which you can't turn round and sell your newly acquired shares. Without knowing what restrictions apply for MS employees, I'd guess that they have to wait between 6 months and a year before being able to sell them. Who knows where MSFT will be in a year? Analysts are predicting a share price near US$200 for the end of 2000, but I'd say that the current spate of MS bad fortune may be causing some hasty revisions of projections in many funds. But historically, MS has always managed to support its share price well year-on-year. After all, from an employee perspective, there is a certain element of golden handcufts with stock options - you always want them to mature before you leave, but if the share price drops sharply, things don't look so rosy any more. In many ways, a long term drop in the share price of MSFT would be a very strong indicator that the MS "World Domination Plan" (TM) is losing its grip.
Cheers,
Toby
Regarding anti-aliased fonts: There's actually a simple way to do that, but it eats RAM like a wumpus. The simple way is not to use fonts at all, but instead to create pixmaps (e.g. XPM format) that have anti-aliased text. This would only be useful on a system with lots(lots) of RAM and then only useful for largely static content like menus, dialogs etc.
I've used a system (Acorn RISC OS) which makes extensive use of fully-anti-aliased fonts (including sub-pixel alignment both vertically and horizontally) and its solution was both elegant and not a memory hog. Starting with vector-based fonts (I'll call them outline fonts) which additionally had hint lines so that thin parts of the letters were not lost at small font sizes, it built pixmaps on the fly and cached them in a pre-allocated font cache. As the text was written to the screen, pre-pixmapped fonts were pulled from the cache and plotted direct: where pixmaps had not been calculated, a block (32?) of characters were rendered into the cache from the relevent font and then plotted. Fonts above a user-defined point-size limit were always calculated and plotted direct from the outline font rather than being cached as pixmaps, and vertical and horizontal sub-pixel alignment could be toggled on and off at will. When plotting the anti-aliased fonts, the pixmap was alpha-blended with any background texture present to avoid strange artefacts. For most of the time, 512K was complete overkill for the font-cache size, and since most documents contain maybe 96 characters in two fonts in two font sizes, the pixmaps don't take up as much space as you might expect. If you change the font in the document, the new font will end up in the cache, and the old one may be discarded if the cache is full, or left in place if the space allows.
Anti-aliased fonts make a huge difference to font readability when done properly, and make a significant impact when sub-pixel alignment is enabled for viewing whole pages of text. No more lines of text separated by one extra line of white pixels, or strange staccato problems when faced with a row of ......... I often had friends comment that the text looked like it had been scanned on a high quality scanner from some postscript output, rather than rendered on the fly, and scrolling speeds on a 200MHz machine rarely, if ever, hiccupped except on documents with vast collections of fonts and font sizes.
Cheers,
Toby Haynes
Corel Linux has been out for some time and I'm still bothered that a company like Corel wants to take their programmers off other projects and reinvent a *better* Debian distro.
Corel has had to work hard on a Linux distro because it has had to build one for its ARM-Linux-based Netwinder machines. All that time and experience in building a non-Intel Linux platform distro is reasonably reinvested in getting an Intel-distro put together to expand their platform presence and provide themselves with a well-understood distro on which to test their software. After all, even in these enlightened platform-independent days (I wish...) Linux on x86 is still the most widespread, and any commercial binary-only releases of software (such as CorelDraw 9) will have to appear on the x86 Linux to see reasonable sales. I strongly hope that they will compile up versions for other Linux platforms as well, but I won't hold my breath for those to appear in the local store - I'd expect those to be dispatched from Corel itself on order.
Cheers,
Toby Haynes
To be honest, I don't think it is just the FPS shooters which are attractive to the Linux crowd. In fact, the history of games on Unix in general seems to suggest that almost anything goes, from the phenomenal longevity of Xpilot with its fast multiplayer action, to the Rogue-like collection (most notably Angband and Nethack, along with Omega and others) and of course the many MUDs. In my opinion, the balance seems to be more towrds thinking, strategy games than anything else.
Now, the reasons for this trend so far may well be that vt terminals are fairly universal, and that the games in the past have been developed for universality. Looking at the current state of development in the Linux game web pages, such as LinuxGames and HappyPenguin, I see various hopeful signs that 3D is really picking up (Parsec, Xracer) and the arrival of Xfree86 4.0 will further enhance developments in this area.
One of the weaknesses of the Linux game development is that the big companies are still too wary of getting into producing Linux games. Loki is doing a great job in getting companies to give the go-ahead for porting games. For the foreseeable future I suspect that Loki's expertise in the Linux field will continue to make this an easier route for games to appear on the Linux platform. It will be a while before Linux expertise is widely grounded inside commercial game companies and Linux versions are produced in-house, although obviously id software has no problems in this area!
Cheers,
Toby Haynes
AMD do have a 1GHz Athlon on the market - it's in the Kryotech SuperG machine, retailing at $2200 I think and available now.
Since Kryotech stuck with AMD through the leaner periods until now, it's nice to see their cooling technology get a worthy processor.
Of course, the general media haven't caught up to this, which is odd given that there was a howling crowd of posters in one of the talkbacks to an article on ZDNet looking at the race to 1GHz all wondering why the Kryotech machines aren't being more widely publicized. No wonder I keep bumping into conspiracy theories about Intel 'warning off' OEMs and PC builders... :-)
Cheers,
Toby Haynes
With so many sites now offering dynamic, up-to-the-minute information, merely caching the contents of a page at a particular moment in time is at best only catching glimpses of these pages, and at worst leads to misleading search results when querying the search database. It strikes me that with these sorts of site, where the content of the pages is changing so rapidly, something more objective is needed.
For example, the first clues about the state of flux of a particular page can be obtained by diff'ing the page against the previous copy held in memory. If the page is simply having extra items added in periodically, such as a FAQ, then the diffs will generally show that there is more being added than taken away, and the traditional snapshot method employed by search engines is fine. However, if the page is wildly different in the majority of its content (such as the slashdot main page), there is practically no point in making a copy of the page for indexing purposes. A much better solution is attempt to build a keyword database automatically for this page by lexical analysis of the text - even a '100-most-common-words' list (with 'the', 'and', 'is' etc filtered out) would be an improvement on the current situation. As repeated visits build up, this keyword list will refine itself and actually provide some reasonable pointers to the material likely to be found at the site.
With any large database, particularly when you get to the stage where GB's of information are consider small fry, the need for efficient data mining and generation of useful indices becomes increasingly important. Database technologies are looking forward to a time when there will be a need for Petabyte and Exabyte storage and retrieval, and effective distillation of a web page's information, rather than a word-for-word verbatim cache will be the only answer.
Cheers,
Toby Haynes
Jon Katz raises all sorts of ethical worries surrounding the Human Genome project, but I feel he misses the point.
The Human Genome Project sets out to *map* the Human Genome - it does not set out to fully understand it. What it does give is a wealth of reference material to help researchers fighting disease and disability try and identify common traits in their subjects.
The Human Genome Project is *not* gene-therapy. It is *not* the new Frankenstein. It is not even being able to choose the appearance of your offspring. Too often the media ends up mistaking information about something as the means to actually do something - a bit like the difference between knowing that sufficient plutonium in the right place can go critical and actually trying to make an effective nuclear weapon. Witness the attempts made by governments all over the world to get a nuclear program going, and the number of successes. The theory is well understood - the practice is more difficult.
Will the Human Genome project change our lives? Possibly. But only as the result of building on the database of information it gives us. The opportunity to allow the development of therapies to improve the quality of life for many people should not be missed. I agree that there are many applications of this information that are ethically-questionable, or morally repugnant, but the Human Genome Project is not the application - it is merely the reference.
Maybe there should be a wider debate on the ethics of genetics, and genetic applications. It would be welcomed to see an informed and intelligent discussion, but with the current media's fixation with hyperbolae and sensation I fear that the important issues will once again get buried beneath an avalanche of "Frankenstein Foods" and similar headlines. Maybe that is an excessively cynical view but I repeatedly see important issues obscured by headline-grabbing stories running on people's fears. In this technological age, the public understanding of science is too often blurred and confused by scare stories in the general press - oh for a day when people are presented with unbiased information with which to make up their own minds.
Cheers,
Toby Haynes