Re:*nix has this stupid fixation with case too
on
ZDNet Reviews KOffice
·
· Score: 2
That sounds like a good reason to use a unicode file system, and is completely orthogonal to filenames that are case sensitive.
Yeah, a unicode file system sounds great, but my point is that it has everything to do with case sensitivity. Case folding of unicode text is complicated and resource intensive. I don't want my kernel doing it.
It's a pain in the @$$ to have to manually edit 5,000 source files.
[Last
time I wrote this, it was Flamebait, so I'll try to be more careful.]
Yes, it is generally agreed that modern encryption algorithms can hide data
with virtually perfect security. But this alone is not relevant, as long as
the government can detect the use of these algorithms.
All the government has to do to nail your "Terrorist Tim" is observe that he
is using encryption, and check for the existance of a matching escrowed key.
Presumably, any key escrow system would allow for verification that a
message was encrypted using an escrowed key, without actually retrieving the
key or decrypting the message. Thus, it is entirely conceivable to me that
the government could enforce the use of key escrow: Whenever they see
encrypted traffic that does not use an escrowed key, they trace the user via
the ISP and prosecute him. And maybe they drop the connection, so you can't
even get one message through then hide.
So, anyone who wants Internet privacy under this regime must hide the fact
that they are hiding data. But, you say, there's a whole field dedicated to
this end, called steganography, so the goverment loses again. While
steganography is exciting and promising, it's not the knock-down argument
that you seem to think.
First, I agree that it is easy to covertly communicate a small amount of
information to someone with whom you have prepared ahead of time. Any
simple system of code words or similar is probably secure for a brief
message or two. But,...
People need to communicate more than a few messages on a predetermined
subject. A naive system will not stand up to statistical analysis of many
messages. For example, you might think that coding messages in the first
characters of each word would be undetectable. Hardly--just look for
anomalies in the letter frequencies of the first letters.
People need to communicate without having arranged a system beforehand.
Even serious steganography (at least the systems I know about and can
imagine) requires a shared secret, implying major challenges in key
exchange. In the age of public keys (now the lynchpin of virtually all
secure communication), we forget about what an enormous breakthrough
asymmetric cryptography was.
Even serious steganography may be detectable! Just as the
government can monitor for non-escrowed keys, they can monitor for any
steganography system that they have broken. It is currently not known
whether undetectable steganography can be developed.
Steganography does not have the infrastructure, either in software or in
familiarity and understanding, that encryption has. We all know that
quality of implementation and good practices are as important as
mathematical strength in the successful use of cryptography. Thus, people
need to have software they can use and an understanding of do's and don't's.
At least, it will take some time before steganography reaches the level of
encryption in these regards.
(In the above, you may substitute "terrorists" for "people".)
The point: not that the government should or will do this; but that if they
decide to do it, it is not futile! It really could (in addition to destroying the privacy of lawful citizens) slow down terrorist
communications (assuming that terrorists use the Internet, which people seem
to think they do). So we need a better argument against it than "this is
stupid, it can't work".
To all the people who say, "Give me freedom or give me death," this is
the time to make your choice.
Well, it's always time to make that choice, and I continue to be comfortable
with the level of risk I assume for my freedom. The chance of being killed
by a terror attack is very low, really. Much lower than the chance of dying
in a car accident.
With any trade-off, you can't just say "more of this requires less of that",
you have to quantify (if roughly) how much of that you have to give to get
how much of this. I recently heard an interview with (I think) Defense
Secretary Rumsfeld. He was asked whether proposed security measures would
improve our defenses against terrorism. Politically, he's obviously
supposed to answer yes, but he hedged. He said, basically, that terrorists
have all the advantages. They operate in a decentralized manner. They
don't require many resources. They can attack at any time and place. He
all but acknowledged that any restrictions that we would be prepared to
accept would make us only marginally safer.
Moreover, all of the emotionally charged discussion about about the balance
between liberty and security distracts us from the ways we can increase
security that have nothing to do with restricting individual liberty. For
example:
Armed guards on airplanes (with non-airplane piercing bullets).
Cockpits inaccessible from the cabin.
Remote override of airplane controls.
Targeted retaliation against terrorists.
Reduction of US aggression abroad (it takes a lot of hate to go through
what the terrorists did).
Increased awareness and vigilance (the terrorists reportedly made many
slips before the attacks; if they had been followed up more vigorously,
things may have been different).
(It's often the case that we fixate unproductively on dualities, when
thinking outside the either-or box reveals the best solutions. For example,
ease-of-use versus functionality in computer software.)
Re:*nix has this stupid fixation with case too
on
ZDNet Reviews KOffice
·
· Score: 3, Insightful
Can anyone give a GOOD reason why the heck you want a file system that is case sensitive?
Think internationalization. Presumably the user should be able to name files in his own language. But Unicode case normalization is expensive and complicated.
(Of course, you could say that case insensitivity need only apply to ASCII, but then you get called a narrow-minded Anglo-centrist.)
Think access control, and all of the bugs that arise when there is more than one name for a resource. Granted, this is probably a software quality problem, but it is reality.
Overally, case sensitive is much simpler engineering-wise. "Normal users" shouldn't have to type exact filenames anyway, they should be using file dialogs or case-insensitive search tools or something.
If I where to send an e-mail that something like this:
Hi George, how's the family? We're doing great over here, Lisa just gave birth to a baby boy, 6 lbs. We're planning on visiting New York
September 12th, and hope we can see before heading home. Will you be in the area? Maybe we can get together for lunch.
Would you know that the sender was REALLY telling the reader to set off a fire bomb(baby boy), approx. 6lbs in weight charge, September 12th at?
You're right that you can get a few important messages on a pretedermined subject through undetected. But try expanding that scheme to wide-scale use. You get into all the problems of key exchange, but worse, since you're not using a key per se but a secret algorithm, which is much bigger to communicate. And, you start to become vulnerable to statistical attacks: the enemy notices that you use some works with unusual frequencies.
If someone wants to hide information, they will, period.
The history of cryptography has shown that the seemingly simple goal of transmitting hidden information is actually really, really hard. The suggestion that if the government outlaws the well known digital privacy schemes, people will come up with others just as good, is naive. It's the same reasoning that says that secret encryption algorithms should be more secure than public algorithms. It grossly underestimates the techniques available to detect and break poorly designed systems.
If the author of OutGuess can detect most steganography, I would not feel at all secure using your "hide the encrypted message in an executable" trick.
And you might look at Stegdetect, by the author of OutGuess. He claims to detect many other popular steganography techniques. The feds throw stegdetect onto carnivore, and you can expect using steganography to earn you one of those unpleasant visits.
Steganography is a long, long way from offering the practical security of encryption. Is it really possible to create a system that is undetectable even if the algorithm is public? Nobody's sure yet. Do the bad guys have the means to create their own effective algorithms and keep them secret? Questionable. Can they use a stego system correctly on a wide scale? Unlikely at present, since there is no popular, easy (for non-technical users) software, nor is there the widespread understanding of how to use stego that there is about crypto (these things do matter when it comes to the successful implementation of any security scheme).
The point is, the government can (by imposing on everyone's liberty) effectively stop criminals from communicating privately. Therefore, we need to come up with a better argument than "it won't work", in order to prevent it.
Criminals, on the other hand, will continue to use widely available crypto packages with no back door and will still be able to transmit messages without threat of law enforcement decrypting them.
Think harder: With carnivore, the government sees all traffic. They see crypto they can't break, they trace it with help from the ISP, they pay someone a not-so-friendly visit.
Please stop convincing yourself it can't work. It can work, and pretending otherwise will only make it more likely.
It sounds like your boss is saddled with the responsibility of being a
leader, and you have the role of a worker. For that reason *alone* it would
be well worth your time, money not to stick your opinionated nose in where
it doesn't belong.
Maybe you've worked in bad situations, but this is completely untrue in a
good company. Any reasonable manager (insert snide question about the
compatiblity of management and reason) knows that the people under him are
intelligent (if they are) and have a perspective that he does not, so they
will often see things that he has missed. If you took a poll, I'm sure that
most managers would wish for workers to participate more in leadership.
Always trust that a business will do the thing that makes it the most
money.
Fundamentally ludicrous: this would imply that companies don't
make mistakes. But a cursory examination of companies, or experience at any
company, shows that companies screw up with regularity. It's sometimes even
clear to observers outside of management that a company is screwing up.
Companies should be grateful for any input from employees that will allow
them to screw up less often.
Sometimes the study says "cofee bad", the next day another one claims "coffe good". With so many divergent results, how can one still trust any of those studies?
Did you actually read the studies? Or did you just read the sensational news headline? I'm fairly confident that none of the studies said "coffee bad" or "coffee good".
You know this, of course, yet you still find it funny to mock the science. This anti-intellectual attitude is the reason that science coverage is so brainless.
If you keep aiming where Microsoft has already been, then your opportunities will be in China.
It sounds nice like a nice tack: provide minimal Microsoft compatibility, while focusing on some vaguely suggested (notice how he avoids any specific discussion of what Sun should do with StarOffice) need that Microsoft doesn't address. What he doesn't get is that there is no such thing as "minimal Microsoft compatibility". This is why the life of an alternative office suite is so miserable.
Let's start with what most people agree on by now: you need to be able to read Office documents that people send you. (Forget for now about creating your own documents, and editing documents that people send you.) According to the article, you just say the magic words "open XML format", wave your wand, and your need for MS Office vanishes in a puff of smoke.
People who say that seem to think you can represent a Word document in a souped-up version of DocBook. Not even close. For starters, there's OLE. This alone is an extremely complicated data model that must be entirely replicated. Not to mention that you have to support every data format that is commonly embedded into Word documents; "just a Word viewer" is an oxymoron. Next, people put formulas in their embedded Excel documents, so you have to clone the scripting language, along with all of the zillions of functions provided. People put macros in their Word documents too, which require in addition to the scripting language a document model that is exactly like Word's. Plus any feature that can be accessed by macros (which I'm guessing is most of them). Oh, these macros might alter the document, so don't think you were going to get away with a read-only model. Compared to all this, emulating the UI is child's play, so to write a Word viewer, you may as well write MS Office.
Basically, Microsoft adds tons of features to Office, and people find the craziest ways to use them, so you have to support every damn one in order to provide "minimal Microsoft compatibility". Anyone who doesn't think it's that bad, probably hasn't worked in a typical business environment.
The alternate notion that people can keep using MS Office for "the full range of functionality in Office", and use StarOffice for the vaguely suggested something else, is just as broken for an even simpler reason: most people don't want to learn more programs.
So maybe China (plus some smaller markets here, like students) is the best Sun can hope for. In a few decades, that may not look like such a bad thing.
You can't learn to write good software by talking and reading about it in
general terms. The best way to learn to write good software is the same as
the best way to learn almost anything: practice, and get feedback from
better programmers. Even though I think I understand where this guy's
coming from, and I basically agree with him, I think his essay is basically
pointless. I think the reason that so many of the posts here are critical
is exactly that he is not specific enough to sway anyone who doesn't already
look at the issue the way he does.
To the extent that we do understand how to write beautiful software, we don't
stand up for it. Although many posters make the obvious point that we are
under pressure to get software done, this is not sufficient excuse for
ignoring the values espoused in the essay, for two reasons: One, software
that merely provides the required functionality may let the user get his job
done, but typically requires him to learn the concepts used in the program
and put up with its ideosynchracies, because the program doesn't work in
terms of the concepts he already uses. This frustrates and alienates the
user, and ultimately contributes to the low regard people have for
computers. Look around you--almost every object pays some passable respect
to usability and form. We shouldn't make (and sell) stuff that's ugly,
cumbersome, confusing, and surprising, because it's dehumanizing.
Two, software that is not conceptually clean is hard to extend. People
often talk about maintainability, but it rarely gets priority during
implementation. Why did Netscape's browser finally lose? Not because they
didn't have good ideas for new features, but because it was internally such
a mess that they couldn't improve it fast enough. This is not uncommon.
So, even when we feel the very necessary pressure to get our code out the
door, we need to push back in order to give more attention to beauty. We will benefit.
disable font antialiasing for font sizes in the range from 8 to 14 pt.
This is exactly right for text widgets and such, but may be a problem for programs that do more sophisticated text layout. If glyphs do not fall on pixel boundaries, anti-aliasing can be a huge win, because forcing glyphs to pixel boundaries can completely screw up glyph spacing. For example, try running xpdf on most any PDF document, in non-anti-aliasing mode.
Does X anti-aliasing have any support for disabling anti-aliasing for "widget text" but enabling it for "WYSIWYG text"? Can programs that need WYSIWYG at least override the default? Italicized, rotated, and magnified text have similar issues. It's not just about point size.
Of course I'm already operating at reasonably high resolution to start with, so there's going to be somewhat less room for improvement through anti-aliasing
Actually, it's rather the opposite: at low resolutions, anti-aliasing is usually less desirable. When the width of a stroke is around a single pixel, a grey pixel stands out in a big way, making the glyph look fuzzy. If glyphs are pixel-aligned (ie, they start and end on pixel boundaries) and upright (not italicized or rotated), a non-anti-aliased, hand-hinted font is much cleaner. (It follows eg that word-processing software should favor magnification levels such that glyphs have integral pixel width and hand-hinting, and fudge a little to put glyphs on pixel boundaries.)
At higher resolutions, there are simply more pixels to play with, and a few grey pixels blend in nicely. 75 dpi versus 100 dpi doesn't make a huge difference, but when we get 300 dpi screens, we'll wonder how we ever put up with today's blocky text.
Today Ben Stein posted an interesting article on thestreet.com, dissecting the myth about the high longterm yields of the stock market.
Dammit, you made me fill out that stupid form just to read a very confused article. Stein purports to refute that that "total return for stocks almost always has eclipsed that for bonds"; but all he demonstrates is that over some periods, stocks (as a whole) have done badly. Ie, stocks are volatile--duh! He says, "To get to the calculation that stocks 'always' outperform bonds and cash, you have to choose your start and stop points selectively.", and "proves" it by picking his counterexamples selectively!
That article was perfect piece of FUD. At least I can take it as a reminder that you don't get better information by filling out sign-up forms.
Actually, we're better off using the same standard for different uses, wherever possible. Do you want to go back to TCP/IP, IPX, and NetBEUI on every LAN? Ethernet and token ring? They all have different uses, but they're close enough that we should just pick one pretending that it will work in all situations, then make the best of it.
Bluetooth and 802.11 are clearly in this situation, IMO. The main difference between them: one is for near and one for far. This makes sense by strict engineering standards, but in the big picture it's a detail. If 802.11 becomes the standard, we'll make it scale down to "near". Not to mention (as did another poster), what do I do if I'm "in between"? There are other parts to Bluetooth, but nothing that can't be layered on top of another network (in the Internet tradition of "dumb network, smart endpoints").
tension continues to grow between the extreme free speech faction and the more moderate folks.
Ulrich is actually a pretty staunch defender of software freedom. I think this is a political and personality conflict, more than a difference in ideology.
But then, Ulrich is quite inscrutable, so I don't claim to speak for him.
Second. If you're in the Boston area, there's a shop called Back Care Basics that sells them. Very hard to find in the US. Expensive, but worth trying out.
good interview with stallman except his dance around the communism question.
Why do you insist that he "danced"? RMS has made clear for years that his movement has nothing to do with communism. Can you not take the man at his word?
Alot of what the FSF and stallman yell about is common to utopian communism.
In the same sense that the spirit of sharing and cooperation in general is common to utopian comumnism. Does it surprise you that many people consider sharing and cooperation wonderful, but loathe the lack of personal economic freedom and concentration of power implied by communism?
Please see my other message in this thread for more.
Socialism: Any of various theories or systems of social organization in which the means of producing and
distributing goods is owned collectively or by a centralized government that often plans and controls the economy.
Thank you for a definition! Often when I discuss this, there is no agreed upon definition, and since I'm not an expert in socialism, I hesitate to provide my own.
That said: GNU does not anywhere propose a "system of social organization". Nor does it talk about collective ownership; indeed RMS emphasizes "Our emphasis is on freedom,
decentralization, and voluntary cooperation" (from the interview). There may be similarities, but the core ideas of socialism are not in GNU, and vice versa.
On the other hand, consider all the flattering things RMS says about America and the american economic system:
'As in "free enterprise" and "free speech", the "free" in "free software" refers to freedom' (from The GNU GPL and the American Way.
It is plain to any person who actually reads RMS: GNU is not about communism or socialism! Neo-socialists: please do RMS the courtesy of not adopting him into your cause.
Simply put, the EFF needs money... if it is to give large enough bribes to the Plutocrats who inhabit your public
offices.
Your message isn't clear enough for me to understand the overall point, but this statement at least is plainly offensive. What does the EFF do that could be construed as bribery?
You might find this overview enlightening. Which of the methods mentioned therein do you find dishonorable?
Or do you just dislike the fact that the EFF seeks change via courts instead of via populism? An institution with a 200 year history of defending civil liberties seems like a good place to turn to me (and the EFF's track record bears this out).
If you think the system is corrupt just because litigation costs money, I don't think you have a good idea of how the world works.
Richard Stallman made a suggestion that is similar to yours, but very characteristically RMS.
I have a suggestion. If I were to suggest totally
boycotting movies, I think people would ignore that
suggestion. They might consider it too radical. So I would like
to make a slightly different suggestion which comes to almost
the same thing in the end, and that is, don't go to a movie
unless you have some substantial reason to think it's good.
Now this will lead in practice to almost the same result as a
total boycott of Hollywood movies. In extension, it's almost
the same but, in intention, it's very different. Now I've noticed
that many people go to movies for reasons that have nothing
to do with whether they think the movies are good. So if you
change that, if you only go to a movie when you have some
substantial reason to think it's good, you'll take away a lot of
their money.
Very like RMS to make a simple observation about human nature, and base on it a proposal that seems at once perfectly natural and hopelessly naive. This seems a paradox only until you realize that RMS works on the time scale of a lifetime. He's already demonstrated that on that scale, the hopeless becomes conceivable. When you step back far enough, he begins to look downright pragmatic.
Observe that your suggestion requires one to reevaluate every purchase in terms of an artificial $5, while his merely to reconsider based on a simple criterion (that I ought to have considered in the first place). I'm not actually commenting on which idea is more effective--I really don't know. But his has an appeal to me that yours lacks.
in case your email has never been revealed anywhere on the net, you can use cgi or php scripts that email you.
Be careful!
Your example demonstrates every mistake it possibly could. One, it requires putting your email address in the HTML, where a spammer could find it. Two, it does not appear to restrict the recipient, meaning it is effectively an open relay. Three, there is no indication that it performs effective logging, meaning it is effectively an anonymous open relay.
Not to mention that any programmer so thoughtless probably didn't think much about security, so you may be creating a new vulnerability without solving the old one.
Slashdot Mirror
Yeah, a unicode file system sounds great, but my point is that it has everything to do with case sensitivity. Case folding of unicode text is complicated and resource intensive. I don't want my kernel doing it.
It's a pain in the @$$ to have to manually edit 5,000 source files.
Yeah, sucks for you, sorry :-)
Yes, it is generally agreed that modern encryption algorithms can hide data with virtually perfect security. But this alone is not relevant, as long as the government can detect the use of these algorithms.
All the government has to do to nail your "Terrorist Tim" is observe that he is using encryption, and check for the existance of a matching escrowed key. Presumably, any key escrow system would allow for verification that a message was encrypted using an escrowed key, without actually retrieving the key or decrypting the message. Thus, it is entirely conceivable to me that the government could enforce the use of key escrow: Whenever they see encrypted traffic that does not use an escrowed key, they trace the user via the ISP and prosecute him. And maybe they drop the connection, so you can't even get one message through then hide.
So, anyone who wants Internet privacy under this regime must hide the fact that they are hiding data. But, you say, there's a whole field dedicated to this end, called steganography, so the goverment loses again. While steganography is exciting and promising, it's not the knock-down argument that you seem to think.
First, I agree that it is easy to covertly communicate a small amount of information to someone with whom you have prepared ahead of time. Any simple system of code words or similar is probably secure for a brief message or two. But, ...
(In the above, you may substitute "terrorists" for "people".)
The point: not that the government should or will do this; but that if they decide to do it, it is not futile! It really could (in addition to destroying the privacy of lawful citizens) slow down terrorist communications (assuming that terrorists use the Internet, which people seem to think they do). So we need a better argument against it than "this is stupid, it can't work".
Well, it's always time to make that choice, and I continue to be comfortable with the level of risk I assume for my freedom. The chance of being killed by a terror attack is very low, really. Much lower than the chance of dying in a car accident.
With any trade-off, you can't just say "more of this requires less of that", you have to quantify (if roughly) how much of that you have to give to get how much of this. I recently heard an interview with (I think) Defense Secretary Rumsfeld. He was asked whether proposed security measures would improve our defenses against terrorism. Politically, he's obviously supposed to answer yes, but he hedged. He said, basically, that terrorists have all the advantages. They operate in a decentralized manner. They don't require many resources. They can attack at any time and place. He all but acknowledged that any restrictions that we would be prepared to accept would make us only marginally safer.
Moreover, all of the emotionally charged discussion about about the balance between liberty and security distracts us from the ways we can increase security that have nothing to do with restricting individual liberty. For example:
(It's often the case that we fixate unproductively on dualities, when thinking outside the either-or box reveals the best solutions. For example, ease-of-use versus functionality in computer software.)
Think internationalization. Presumably the user should be able to name files in his own language. But Unicode case normalization is expensive and complicated. (Of course, you could say that case insensitivity need only apply to ASCII, but then you get called a narrow-minded Anglo-centrist.)
Think access control, and all of the bugs that arise when there is more than one name for a resource. Granted, this is probably a software quality problem, but it is reality.
Overally, case sensitive is much simpler engineering-wise. "Normal users" shouldn't have to type exact filenames anyway, they should be using file dialogs or case-insensitive search tools or something.
Hi George, how's the family? We're doing great over here, Lisa just gave birth to a baby boy, 6 lbs. We're planning on visiting New York September 12th, and hope we can see before heading home. Will you be in the area? Maybe we can get together for lunch.
Would you know that the sender was REALLY telling the reader to set off a fire bomb(baby boy), approx. 6lbs in weight charge, September 12th at?
You're right that you can get a few important messages on a pretedermined subject through undetected. But try expanding that scheme to wide-scale use. You get into all the problems of key exchange, but worse, since you're not using a key per se but a secret algorithm, which is much bigger to communicate. And, you start to become vulnerable to statistical attacks: the enemy notices that you use some works with unusual frequencies.
The history of cryptography has shown that the seemingly simple goal of transmitting hidden information is actually really, really hard. The suggestion that if the government outlaws the well known digital privacy schemes, people will come up with others just as good, is naive. It's the same reasoning that says that secret encryption algorithms should be more secure than public algorithms. It grossly underestimates the techniques available to detect and break poorly designed systems.
If the author of OutGuess can detect most steganography, I would not feel at all secure using your "hide the encrypted message in an executable" trick.
And you might look at Stegdetect, by the author of OutGuess. He claims to detect many other popular steganography techniques. The feds throw stegdetect onto carnivore, and you can expect using steganography to earn you one of those unpleasant visits.
Steganography is a long, long way from offering the practical security of encryption. Is it really possible to create a system that is undetectable even if the algorithm is public? Nobody's sure yet. Do the bad guys have the means to create their own effective algorithms and keep them secret? Questionable. Can they use a stego system correctly on a wide scale? Unlikely at present, since there is no popular, easy (for non-technical users) software, nor is there the widespread understanding of how to use stego that there is about crypto (these things do matter when it comes to the successful implementation of any security scheme).
The point is, the government can (by imposing on everyone's liberty) effectively stop criminals from communicating privately. Therefore, we need to come up with a better argument than "it won't work", in order to prevent it.
Think harder: With carnivore, the government sees all traffic. They see crypto they can't break, they trace it with help from the ISP, they pay someone a not-so-friendly visit.
Please stop convincing yourself it can't work. It can work, and pretending otherwise will only make it more likely.
Maybe you've worked in bad situations, but this is completely untrue in a good company. Any reasonable manager (insert snide question about the compatiblity of management and reason) knows that the people under him are intelligent (if they are) and have a perspective that he does not, so they will often see things that he has missed. If you took a poll, I'm sure that most managers would wish for workers to participate more in leadership.
Always trust that a business will do the thing that makes it the most money.
Fundamentally ludicrous: this would imply that companies don't make mistakes. But a cursory examination of companies, or experience at any company, shows that companies screw up with regularity. It's sometimes even clear to observers outside of management that a company is screwing up. Companies should be grateful for any input from employees that will allow them to screw up less often.
Did you actually read the studies? Or did you just read the sensational news headline? I'm fairly confident that none of the studies said "coffee bad" or "coffee good".
You know this, of course, yet you still find it funny to mock the science. This anti-intellectual attitude is the reason that science coverage is so brainless.
It sounds nice like a nice tack: provide minimal Microsoft compatibility, while focusing on some vaguely suggested (notice how he avoids any specific discussion of what Sun should do with StarOffice) need that Microsoft doesn't address. What he doesn't get is that there is no such thing as "minimal Microsoft compatibility". This is why the life of an alternative office suite is so miserable.
Let's start with what most people agree on by now: you need to be able to read Office documents that people send you. (Forget for now about creating your own documents, and editing documents that people send you.) According to the article, you just say the magic words "open XML format", wave your wand, and your need for MS Office vanishes in a puff of smoke.
People who say that seem to think you can represent a Word document in a souped-up version of DocBook. Not even close. For starters, there's OLE. This alone is an extremely complicated data model that must be entirely replicated. Not to mention that you have to support every data format that is commonly embedded into Word documents; "just a Word viewer" is an oxymoron. Next, people put formulas in their embedded Excel documents, so you have to clone the scripting language, along with all of the zillions of functions provided. People put macros in their Word documents too, which require in addition to the scripting language a document model that is exactly like Word's. Plus any feature that can be accessed by macros (which I'm guessing is most of them). Oh, these macros might alter the document, so don't think you were going to get away with a read-only model. Compared to all this, emulating the UI is child's play, so to write a Word viewer, you may as well write MS Office.
Basically, Microsoft adds tons of features to Office, and people find the craziest ways to use them, so you have to support every damn one in order to provide "minimal Microsoft compatibility". Anyone who doesn't think it's that bad, probably hasn't worked in a typical business environment.
The alternate notion that people can keep using MS Office for "the full range of functionality in Office", and use StarOffice for the vaguely suggested something else, is just as broken for an even simpler reason: most people don't want to learn more programs.
So maybe China (plus some smaller markets here, like students) is the best Sun can hope for. In a few decades, that may not look like such a bad thing.
Two, software that is not conceptually clean is hard to extend. People often talk about maintainability, but it rarely gets priority during implementation. Why did Netscape's browser finally lose? Not because they didn't have good ideas for new features, but because it was internally such a mess that they couldn't improve it fast enough. This is not uncommon.
So, even when we feel the very necessary pressure to get our code out the door, we need to push back in order to give more attention to beauty. We will benefit.
Note to OSNews: Exclusive means that the subjects agreed not to talk to anyone else, not that you're the only ones who bothered to interview them.
Nice post! I showed it to Linus Torvalds and, though it was hard for him, he finally agreed to scrap Linux and work on the HURD. One battle won!
I'm going to talk to Bram Moolenaar next, because I'm pretty sure there was another vi clone before vim.
This is exactly right for text widgets and such, but may be a problem for programs that do more sophisticated text layout. If glyphs do not fall on pixel boundaries, anti-aliasing can be a huge win, because forcing glyphs to pixel boundaries can completely screw up glyph spacing. For example, try running xpdf on most any PDF document, in non-anti-aliasing mode.
Does X anti-aliasing have any support for disabling anti-aliasing for "widget text" but enabling it for "WYSIWYG text"? Can programs that need WYSIWYG at least override the default? Italicized, rotated, and magnified text have similar issues. It's not just about point size.
Actually, it's rather the opposite: at low resolutions, anti-aliasing is usually less desirable. When the width of a stroke is around a single pixel, a grey pixel stands out in a big way, making the glyph look fuzzy. If glyphs are pixel-aligned (ie, they start and end on pixel boundaries) and upright (not italicized or rotated), a non-anti-aliased, hand-hinted font is much cleaner. (It follows eg that word-processing software should favor magnification levels such that glyphs have integral pixel width and hand-hinting, and fudge a little to put glyphs on pixel boundaries.)
At higher resolutions, there are simply more pixels to play with, and a few grey pixels blend in nicely. 75 dpi versus 100 dpi doesn't make a huge difference, but when we get 300 dpi screens, we'll wonder how we ever put up with today's blocky text.
Dammit, you made me fill out that stupid form just to read a very confused article. Stein purports to refute that that "total return for stocks almost always has eclipsed that for bonds"; but all he demonstrates is that over some periods, stocks (as a whole) have done badly. Ie, stocks are volatile--duh! He says, "To get to the calculation that stocks 'always' outperform bonds and cash, you have to choose your start and stop points selectively.", and "proves" it by picking his counterexamples selectively!
That article was perfect piece of FUD. At least I can take it as a reminder that you don't get better information by filling out sign-up forms.
Actually, we're better off using the same standard for different uses, wherever possible. Do you want to go back to TCP/IP, IPX, and NetBEUI on every LAN? Ethernet and token ring? They all have different uses, but they're close enough that we should just pick one pretending that it will work in all situations, then make the best of it.
Bluetooth and 802.11 are clearly in this situation, IMO. The main difference between them: one is for near and one for far. This makes sense by strict engineering standards, but in the big picture it's a detail. If 802.11 becomes the standard, we'll make it scale down to "near". Not to mention (as did another poster), what do I do if I'm "in between"? There are other parts to Bluetooth, but nothing that can't be layered on top of another network (in the Internet tradition of "dumb network, smart endpoints").
Ulrich is actually a pretty staunch defender of software freedom. I think this is a political and personality conflict, more than a difference in ideology.
But then, Ulrich is quite inscrutable, so I don't claim to speak for him.
Second. If you're in the Boston area, there's a shop called Back Care Basics that sells them. Very hard to find in the US. Expensive, but worth trying out.
Why do you insist that he "danced"? RMS has made clear for years that his movement has nothing to do with communism. Can you not take the man at his word?
Alot of what the FSF and stallman yell about is common to utopian communism.
In the same sense that the spirit of sharing and cooperation in general is common to utopian comumnism. Does it surprise you that many people consider sharing and cooperation wonderful, but loathe the lack of personal economic freedom and concentration of power implied by communism?
Please see my other message in this thread for more.
Thank you for a definition! Often when I discuss this, there is no agreed upon definition, and since I'm not an expert in socialism, I hesitate to provide my own.
That said: GNU does not anywhere propose a "system of social organization". Nor does it talk about collective ownership; indeed RMS emphasizes "Our emphasis is on freedom, decentralization, and voluntary cooperation" (from the interview). There may be similarities, but the core ideas of socialism are not in GNU, and vice versa.
On the other hand, consider all the flattering things RMS says about America and the american economic system: 'As in "free enterprise" and "free speech", the "free" in "free software" refers to freedom' (from The GNU GPL and the American Way.
It is plain to any person who actually reads RMS: GNU is not about communism or socialism! Neo-socialists: please do RMS the courtesy of not adopting him into your cause.
Your message isn't clear enough for me to understand the overall point, but this statement at least is plainly offensive. What does the EFF do that could be construed as bribery? You might find this overview enlightening. Which of the methods mentioned therein do you find dishonorable?
Or do you just dislike the fact that the EFF seeks change via courts instead of via populism? An institution with a 200 year history of defending civil liberties seems like a good place to turn to me (and the EFF's track record bears this out). If you think the system is corrupt just because litigation costs money, I don't think you have a good idea of how the world works.
Very like RMS to make a simple observation about human nature, and base on it a proposal that seems at once perfectly natural and hopelessly naive. This seems a paradox only until you realize that RMS works on the time scale of a lifetime. He's already demonstrated that on that scale, the hopeless becomes conceivable. When you step back far enough, he begins to look downright pragmatic.
Observe that your suggestion requires one to reevaluate every purchase in terms of an artificial $5, while his merely to reconsider based on a simple criterion (that I ought to have considered in the first place). I'm not actually commenting on which idea is more effective--I really don't know. But his has an appeal to me that yours lacks.
Be careful! Your example demonstrates every mistake it possibly could. One, it requires putting your email address in the HTML, where a spammer could find it. Two, it does not appear to restrict the recipient, meaning it is effectively an open relay. Three, there is no indication that it performs effective logging, meaning it is effectively an anonymous open relay.
Not to mention that any programmer so thoughtless probably didn't think much about security, so you may be creating a new vulnerability without solving the old one.