Have a look at XML. Standardised file and yet really easy to add new information, no?
Designing a file format so that every application can extract at least the basic information is pretty easy. It isn't even that hard to organise the extensions in a nice heirachy so that certain applications support some more sophisticated features without having to support the entire spec.
I'm inclined to disagree. I guess we'll find out which of us is right over the next 20 or so years:-)
I cannot imagine any system in which the computer can assume all other objects which must be avoided are either computers or stationary. For a start there are things like kids or dogs running across the road. Then there are the changes in the road layout which someone forgot to report. Finally there are the broken computers (bit-flips) and crazy humans deciding to take the fast lane.
The only way I can see it happening is with computers interacting fully with manually controlled cars and that it won't be until long after everyone has computer controlled cars that some lanes become computer exclusive (i.e. fast lanes).
Still, it gets significantly complicated by the ligigeous dangers. If computers are restricted to a single lane and some idiot wanders in then that is their fault. However, if the computer is sharing a lane and some idiot slams on their brakes then I bet they'd try to blame the computer. Given that, maybe you'd get one approach developing throughout the world and another developing in America.
Legacy support... that's what they were smoking...
Back in, gee, was it really 84? they had noted that simply distributing a.EXE was not enough to run an application since it needed all the graphics, sounds, etc. They thought it would be really neat if those were all bundled with the application. So far I agree 100% with them.
The solution (data fork and resource fork) seems a bit crazy now. The.app approach (a folder with a special name = a program) seems much better. But I can see why they were proud of it at the time. After all, what is a folder but a hacky (but easily maintained) way of implementing a resource fork). Plus, the resource fork made it much easier to get POSIX compliance, just enforce all open/creat calls only work on the DATA section. How on earth do you open a folder?
Even ten years later (95 and so), resource forks were paying nice dividends for apple. While PC users clicked on a.txt file only to have it open in some 'default', mac uses had it automatically open in the program which created it. They even had the icon automatically set correctly, the language, etc. Not that many people bothered to give their report documents a different icon to their memos.
Later linux caught up (by and large) with clever use of/etc/magic. Windows simply gave up on.txt AFAICT. Now that we have/etc/magic i'm not convinced the mac approach is worthwhile anymore but with all those old mac programs and old mac conventions -- there really was no other way of handling them. I'm guessing I would have done the same (of course, I'd have got cp working with them right from the start!)
cp etc work perfectly for unixy work. I can edit code, compile it, run it, etc. all just fine. If you treat the machine just like a unix box then you'll never run into trouble with the old cp, mv, etc.
Apple bolted macos onto a unix shell. That bolting is (pretty much) seamless from a user's perspective, but if you're in the shell using unix tools then the bolting is more obvious. For example, you could copy applications between unix computers using something like rsync -avz --delete -e ssh/usr/local:/usr/
Copying applications on macos using the same command with/Applications (the macos programs) won't work well. However, copying/sw (the unix programs) will work just fine since it is 'unixy'.
Do you follow? MacOS users who don't even know their computer has a command prompt will never have any problems, and unix users who largely ignore the GUI will never have any problems. The people who will have problems are those trying to tie the two, writing shell scripts to automate GUI operations, etc. I guess these are the hardcore apple geeks who want to get the most out of their system, and the unix admins who have been told that since macos is unix they are now in charge of maintaining the macs.
Now, that's still bad form by Apple; nobody should have to run a special copy command since you're copying a different type of file. After all, in linux you can make a copy of/dev/zero just by typing cp/dev/zero/dev/zeros. What's that? You can't? You have to type cpio/dev/zero? Well, hopefully you get the point about the behaviour not coming up much, and those few people who are affected know the workarounds.
Yeah, it sounds pretty slack of apple but in reality it hardly ever affected me. You do mac things, or you do unixy things, but I rarely did unixy things on mac software -- perhaps the closest I got was mixing TeXshop/preview and tetex/make.
Similarly, apple's NFS server doesn't support named pipes. Sounds crazy, but realistically how often do you use named pipes via NFS?
Er... You're aware that PNG is 24 bit rather than 16, right? And that float vs integer doesn't make a lot of sense when you're measuring in bits... Have a look at the PNG spec sometime, e.g. the filter specification. I think you might be surprised.
I wouldn't call TIFF a poor man's PNG, but it is certainly showing its age. Oh, and your finishing comment about TIFF sadly applies just as well to PNG...
I suppose they could really confuse people by creating a new version of the standard which is neither open nor backwards compatible, but I'd be surprised if they were stupid enough to do that.
John Dvorak may be more of a journalist than say Rob Enderle or Laura Didio, but the guy is a nutter. Have a look at his comments on the current iMac: "The design is hardly inspirational. In fact, if you put two headlamps on it and a metal sun visor over its "windshield," it would be reminiscent of a 1954 DeSoto." Or perhaps his opinion that Linux would die as soon as MS released a distro http://www.pcmag.com/article2/0,1759,1768170,00.as p
I would trust a random guy on slashdot much more than I'd trust Dvorak's insights...
If cellphones really were effective in taking out planes, you'd see them banned as an anti-terrorism measure. That they're not strongly implies to me that cellphones have only irritating effects on flying.
KDE added support for this some time ago (try hitting 'lock screen'). Alternatively, some distros start KDM(/GDM/XDM) automatically on F7, F8, F9 -- see/etc/X11/xdm/settings or somesuch...
But you're right, it isn't as easy as it should be.
I guess they're happy with non-geeks hitting logout and relying on KDE to correctly restore your session when you get back (which it does a moderately good job of)
The Pentium M uses a lot less power than the sempron (12W vs 65W off the top of my head). However, there are no Pentium-M boards suitable for servers and only a tiny number suitable for desktops. Putting together a Pentium-M based desktop costs quite a bit more than going the A64 route too -- those mobile chips are expensive.
The most effective approach I've seen to this is to define the ability to consent in terms of the age (and therefore 'power') difference. That is, a fourteen year old can consent to having sex with a fifteen year old, but not with an eighteen year old.
However, this approach makes particularly liberal people uncomfortable since they don't like the idea that you can legally have sex with some people but not with others (where the others can legally have sex with some people). It also makes particularly conservative people uncomfortable since they don't like the idea that their fourteen year old daughter can legally have sex.
Since it isn't getting picked up by either the liberals or the conservatives, I can't see the US adopting it. But that's politics for you...
I remember an early version of yacc was GPLed (GPL v1 from memory). That included the boiler-plate code which it inserted into your program, which means every parser built using it was using GPLed code.
Just as in this case, that was an accident and after much apologising the yacc people relicenced the boilerplate code to public domain. It simply is not the intention of free software developers to 'infect' other programs.
I would anticipate a very similar solution in this case. The licence for the fonts just needs a minor tweak explicitly allowing their inclusion and everything goes back to how you'd expect.
It's legalesse. People stuff up occasionally. When they notice, they fix it and move on.
Well, I've only used two G5 machines; my wife's iMac at 1.8GHz and what was a decent mainframe a couple years ago with 16GB of RAM and very fast SCSI RAID, but only 1GHz.
Comparing them to the athlon is always awkward. but the lowest end athlon (my machine) is 1.8GHz and feels about as fast as my wife's machine. Some things are faster, others slower. The research machine at 1GHz generally outperforms the 1.8GHz machine, at least with what I do on it. So clock-for-clock that's doing pretty well compared to the athlon.
However, these anecdotal comparisons relate to the kind of tests I do (training neural networks, singular value decomposition, lots of hash-table lookups/ram access). For my stuff, intel does awfully, a 1200+ (what is that, 600MHz?) outperforms a 2 GHz intel. No doubt people doing different things would get different figures.
Regardless, my point is that Intel tried to push the MHz myth since that is how they sold chips. AMD has convinced the tech community that the MHz myth really is a myth and so when you hear apple arguing the same thing it is easier to believe than it would be without AMD.
Yah, some software I'm looking at (http://www.open-dent.com/) does this too. Personally I don't like it very much, and I think it discourages a community developing around the software. But, their code, their call and all that.
Try the latest version, it has virtually every feature of acroread and works perfectly.
What's it missing? Rotate -- or does it have that? I forget. And forms, unless you get the very latest CVS version. Oh, and I don't think it supports that document tracking spyware;-)
The load time is because it is loading motif. If you already had those libs running (or with acro 7 if you have that version of GTK running) then load time is much more acceptable -- about 5s slower than xpdf.
Oh, and kpdf has automatic change detection so you don't have to hit R:-P
While this was true, forcing everyone to use xpdf had the huge benefit of strongly pushing xpdf development. Just a few months ago they came out in kpdf which has virtually every feature necessary in a fraction of acrobat's footprint.
However all the code has been passed back to xpdf, so you can expect newer versions of xpdf and gpdf to have all the whizz-bang features in kpdf RSN.
Oh, forms weren't in the kpdf that was released a few months ago. Either run that pdf2wysiwyg script someone posted yesterday, or download the {g,k,x}pdf CVS, or wait a couple months for the next version of kpdf/gpdf/xpdf to be released.
The only really cool feature it is missing is a sophistiticated search -- sadly they modelled search based on acroread rather than on the much more powerful search in apple's 'preview.app'
Does it matter? If people want to come along with wrong ideas then they might miss out for a few years until they stumble apon the truth. What's the damage done to linux? What's the damage done to them? Sure, they've missed out a little bit, and maybe linux missed out on a couple early contributions from them. But overall I think it gives us even more time to make linux even more 'ready for the desktop' (which, compared to OSX, it isn't)
And consistant look and feel is the hallmark of xandros. It is also supplied rather well by linspire. Hell, even RedHat^WFedora value consistancy over features to some extent.
Other distros, such as ubuntu, believe it is more important to provide users with the options. But the point is that if consistancy is your goal, then there are linux distributions available today which do a damn good job of satisfying it.
I don't think this is apple's doing. What has happened is that in tech circles, AMD has got the upper hand and so we take their specs seriously.
The top AMD is, what, 2.6GHz? Compared to intel at 4GHz. We've come to accept that AMD's 2.6GHz is faster than intel's 4GHz, and from there it isn't hard to believe that IBM is approximately equal to AMD per MHz. But I don't think it was apple fanboy posts that demonstrated this, I think it was watching A64 machines run rings around intel.
Incidentially, I was doing some work on a 970 last week and it ran faster than my (32 bit) athlon at twice the clock. I haven't had time to look at why, but I suspect the compiler noticed I was doing stuff with vectors and managed to get it all as a single altivec instruction, but on the athlon it was unable to represent it as a single SSE2 instruction.
Slashdot Mirror
Have a look at XML. Standardised file and yet really easy to add new information, no?
Designing a file format so that every application can extract at least the basic information is pretty easy. It isn't even that hard to organise the extensions in a nice heirachy so that certain applications support some more sophisticated features without having to support the entire spec.
I'm inclined to disagree. I guess we'll find out which of us is right over the next 20 or so years :-)
I cannot imagine any system in which the computer can assume all other objects which must be avoided are either computers or stationary. For a start there are things like kids or dogs running across the road. Then there are the changes in the road layout which someone forgot to report. Finally there are the broken computers (bit-flips) and crazy humans deciding to take the fast lane.
The only way I can see it happening is with computers interacting fully with manually controlled cars and that it won't be until long after everyone has computer controlled cars that some lanes become computer exclusive (i.e. fast lanes).
Still, it gets significantly complicated by the ligigeous dangers. If computers are restricted to a single lane and some idiot wanders in then that is their fault. However, if the computer is sharing a lane and some idiot slams on their brakes then I bet they'd try to blame the computer. Given that, maybe you'd get one approach developing throughout the world and another developing in America.
Legacy support... that's what they were smoking...
.EXE was not enough to run an application since it needed all the graphics, sounds, etc. They thought it would be really neat if those were all bundled with the application. So far I agree 100% with them.
.app approach (a folder with a special name = a program) seems much better. But I can see why they were proud of it at the time. After all, what is a folder but a hacky (but easily maintained) way of implementing a resource fork). Plus, the resource fork made it much easier to get POSIX compliance, just enforce all open/creat calls only work on the DATA section. How on earth do you open a folder?
.txt file only to have it open in some 'default', mac uses had it automatically open in the program which created it. They even had the icon automatically set correctly, the language, etc. Not that many people bothered to give their report documents a different icon to their memos.
/etc/magic. Windows simply gave up on .txt AFAICT. Now that we have /etc/magic i'm not convinced the mac approach is worthwhile anymore but with all those old mac programs and old mac conventions -- there really was no other way of handling them. I'm guessing I would have done the same (of course, I'd have got cp working with them right from the start!)
Back in, gee, was it really 84? they had noted that simply distributing a
The solution (data fork and resource fork) seems a bit crazy now. The
Even ten years later (95 and so), resource forks were paying nice dividends for apple. While PC users clicked on a
Later linux caught up (by and large) with clever use of
cp etc work perfectly for unixy work. I can edit code, compile it, run it, etc. all just fine. If you treat the machine just like a unix box then you'll never run into trouble with the old cp, mv, etc.
/usr/local :/usr/
/Applications (the macos programs) won't work well. However, copying /sw (the unix programs) will work just fine since it is 'unixy'.
/dev/zero just by typing cp /dev/zero /dev/zeros. What's that? You can't? You have to type cpio /dev/zero? Well, hopefully you get the point about the behaviour not coming up much, and those few people who are affected know the workarounds.
Apple bolted macos onto a unix shell. That bolting is (pretty much) seamless from a user's perspective, but if you're in the shell using unix tools then the bolting is more obvious. For example, you could copy applications between unix computers using something like rsync -avz --delete -e ssh
Copying applications on macos using the same command with
Do you follow? MacOS users who don't even know their computer has a command prompt will never have any problems, and unix users who largely ignore the GUI will never have any problems. The people who will have problems are those trying to tie the two, writing shell scripts to automate GUI operations, etc. I guess these are the hardcore apple geeks who want to get the most out of their system, and the unix admins who have been told that since macos is unix they are now in charge of maintaining the macs.
Now, that's still bad form by Apple; nobody should have to run a special copy command since you're copying a different type of file. After all, in linux you can make a copy of
Yeah, it sounds pretty slack of apple but in reality it hardly ever affected me. You do mac things, or you do unixy things, but I rarely did unixy things on mac software -- perhaps the closest I got was mixing TeXshop/preview and tetex/make.
Similarly, apple's NFS server doesn't support named pipes. Sounds crazy, but realistically how often do you use named pipes via NFS?
Er... You're aware that PNG is 24 bit rather than 16, right? And that float vs integer doesn't make a lot of sense when you're measuring in bits... Have a look at the PNG spec sometime, e.g. the filter specification. I think you might be surprised.
I wouldn't call TIFF a poor man's PNG, but it is certainly showing its age. Oh, and your finishing comment about TIFF sadly applies just as well to PNG...
Er, no. That's why standards are published.
I suppose they could really confuse people by creating a new version of the standard which is neither open nor backwards compatible, but I'd be surprised if they were stupid enough to do that.
John Dvorak may be more of a journalist than say Rob Enderle or Laura Didio, but the guy is a nutter. Have a look at his comments on the current iMac: "The design is hardly inspirational. In fact, if you put two headlamps on it and a metal sun visor over its "windshield," it would be reminiscent of a 1954 DeSoto." Or perhaps his opinion that Linux would die as soon as MS released a distro http://www.pcmag.com/article2/0,1759,1768170,00.as p
I would trust a random guy on slashdot much more than I'd trust Dvorak's insights...
You missed the testing phase, which is by far the longest (and therefore slows the rest down).
If cellphones really were effective in taking out planes, you'd see them banned as an anti-terrorism measure. That they're not strongly implies to me that cellphones have only irritating effects on flying.
KDE added support for this some time ago (try hitting 'lock screen'). Alternatively, some distros start KDM(/GDM/XDM) automatically on F7, F8, F9 -- see /etc/X11/xdm/settings or somesuch...
But you're right, it isn't as easy as it should be.
I guess they're happy with non-geeks hitting logout and relying on KDE to correctly restore your session when you get back (which it does a moderately good job of)
The Pentium M uses a lot less power than the sempron (12W vs 65W off the top of my head). However, there are no Pentium-M boards suitable for servers and only a tiny number suitable for desktops. Putting together a Pentium-M based desktop costs quite a bit more than going the A64 route too -- those mobile chips are expensive.
The most effective approach I've seen to this is to define the ability to consent in terms of the age (and therefore 'power') difference. That is, a fourteen year old can consent to having sex with a fifteen year old, but not with an eighteen year old.
However, this approach makes particularly liberal people uncomfortable since they don't like the idea that you can legally have sex with some people but not with others (where the others can legally have sex with some people). It also makes particularly conservative people uncomfortable since they don't like the idea that their fourteen year old daughter can legally have sex.
Since it isn't getting picked up by either the liberals or the conservatives, I can't see the US adopting it. But that's politics for you...
I remember an early version of yacc was GPLed (GPL v1 from memory). That included the boiler-plate code which it inserted into your program, which means every parser built using it was using GPLed code.
Just as in this case, that was an accident and after much apologising the yacc people relicenced the boilerplate code to public domain. It simply is not the intention of free software developers to 'infect' other programs.
I would anticipate a very similar solution in this case. The licence for the fonts just needs a minor tweak explicitly allowing their inclusion and everything goes back to how you'd expect.
It's legalesse. People stuff up occasionally. When they notice, they fix it and move on.
I bet that at -39 you wouldn't even need a heatsink!
Well, I've only used two G5 machines; my wife's iMac at 1.8GHz and what was a decent mainframe a couple years ago with 16GB of RAM and very fast SCSI RAID, but only 1GHz.
Comparing them to the athlon is always awkward. but the lowest end athlon (my machine) is 1.8GHz and feels about as fast as my wife's machine. Some things are faster, others slower. The research machine at 1GHz generally outperforms the 1.8GHz machine, at least with what I do on it. So clock-for-clock that's doing pretty well compared to the athlon.
However, these anecdotal comparisons relate to the kind of tests I do (training neural networks, singular value decomposition, lots of hash-table lookups/ram access). For my stuff, intel does awfully, a 1200+ (what is that, 600MHz?) outperforms a 2 GHz intel. No doubt people doing different things would get different figures.
Regardless, my point is that Intel tried to push the MHz myth since that is how they sold chips. AMD has convinced the tech community that the MHz myth really is a myth and so when you hear apple arguing the same thing it is easier to believe than it would be without AMD.
Yah, some software I'm looking at (http://www.open-dent.com/) does this too. Personally I don't like it very much, and I think it discourages a community developing around the software. But, their code, their call and all that.
Try the latest version, it has virtually every feature of acroread and works perfectly.
;-)
What's it missing? Rotate -- or does it have that? I forget. And forms, unless you get the very latest CVS version. Oh, and I don't think it supports that document tracking spyware
The load time is because it is loading motif. If you already had those libs running (or with acro 7 if you have that version of GTK running) then load time is much more acceptable -- about 5s slower than xpdf.
:-P
Oh, and kpdf has automatic change detection so you don't have to hit R
While this was true, forcing everyone to use xpdf had the huge benefit of strongly pushing xpdf development. Just a few months ago they came out in kpdf which has virtually every feature necessary in a fraction of acrobat's footprint.
However all the code has been passed back to xpdf, so you can expect newer versions of xpdf and gpdf to have all the whizz-bang features in kpdf RSN.
Oh, forms weren't in the kpdf that was released a few months ago. Either run that pdf2wysiwyg script someone posted yesterday, or download the {g,k,x}pdf CVS, or wait a couple months for the next version of kpdf/gpdf/xpdf to be released.
The only really cool feature it is missing is a sophistiticated search -- sadly they modelled search based on acroread rather than on the much more powerful search in apple's 'preview.app'
Does it matter? If people want to come along with wrong ideas then they might miss out for a few years until they stumble apon the truth. What's the damage done to linux? What's the damage done to them? Sure, they've missed out a little bit, and maybe linux missed out on a couple early contributions from them. But overall I think it gives us even more time to make linux even more 'ready for the desktop' (which, compared to OSX, it isn't)
And consistant look and feel is the hallmark of xandros. It is also supplied rather well by linspire. Hell, even RedHat^WFedora value consistancy over features to some extent.
Other distros, such as ubuntu, believe it is more important to provide users with the options. But the point is that if consistancy is your goal, then there are linux distributions available today which do a damn good job of satisfying it.
I don't think this is apple's doing. What has happened is that in tech circles, AMD has got the upper hand and so we take their specs seriously.
The top AMD is, what, 2.6GHz? Compared to intel at 4GHz. We've come to accept that AMD's 2.6GHz is faster than intel's 4GHz, and from there it isn't hard to believe that IBM is approximately equal to AMD per MHz. But I don't think it was apple fanboy posts that demonstrated this, I think it was watching A64 machines run rings around intel.
Incidentially, I was doing some work on a 970 last week and it ran faster than my (32 bit) athlon at twice the clock. I haven't had time to look at why, but I suspect the compiler noticed I was doing stuff with vectors and managed to get it all as a single altivec instruction, but on the athlon it was unable to represent it as a single SSE2 instruction.
Interesting, thanks :)
MS research comes up with some very cool stuff. Virtually none of it makes its way into Microsoft's products, I have no idea why :(