Actually, I'm I'm not fully convinced that the mbox format cannot be indexed by spotlight. The contact database of address book is also in a single file and it can be searched. If I remember Ars Technica's excellent article on Tiger, there is a trick to do this.
The idea behind launchd is to have one single daemon to replace all daemon that launch other processes, this means initd, crond, inted, etc...
The advantages of this approach are the following:
All those daemon have shared functionality but different code, this means more code to maintain and more security risks (all those daemon have to run at some point as root). Each of the 'old' daemon also has a different file format, which does not make life easier either.
Process launch criteria can be a mix of those offered by the 'old' daemons, crond is based on time, inetd on network connections, initd on boot sequence. Launchd 'understands" all those notions.
The example I was given was of starting a backup task on a laptop. The criteria to launch the task would be something like if last execution was more than X days ago and the computer is not running on battery power and there is a network connection and CPU load has been low for some time then launch...
Launchd is supposed to have 'adapters' to understand the old file formats.
Personally I think this is a good idea, factoring out common functionality and using more reasonable file formats, but of course the old guard will complain that the current set of daemon just works (not on a laptop) and that this was proposed by people who do not understand Unix - who is this Jordan Hubbard anyway?:-)
The privacy problem in Office Documents was mostly that the file contained data that should not have been there. Somehow, if the meta-information extractors can extract more information that the user can by opening the file with the appropriate application, then there is something fishy in your application and is IMHO not a meta data problem per se.
I would be more interested to see if the meta-index respects the permissions (Unix or ACLs), i.e that an user can only see meta-information of files he can read.
I agree, if more articles linked by slashdot were of this quality, I would sure be happy. Then again I'm really disappointed by the discussion here on slashdot, we get a really technical article (stuff for that matters), and people keep bitching about the price of the upgrade, making silly wishes for g5 laptops or OS for intel.
A few months ago, Jordan Hubbard came to CERN to talk about some of the Unix elements of Tiger, and talked about launchd. I think that this is one of the features of Tiger that should be cloned ported to Linux (John Siracusa seems to agree). Having an unified launching mechanism for processes is really something that is needed on Unix, especially for laptops.
You really want to be able to launch processes depending on different triggers and circumstances, like saying at that time, if the machine has been idle for some time and I'm not running on battery power, then launch that process. Yes, you can do hack similar functionality with scripts, but no, this is not convenient or stable.
As people have more and more old machines lying around, having a live CD with a single purpose operating system. I think there are already distributions that transform a PC into a router, but changing an old PC into a firewire hard drive enclosure looks like an interesting idea.
One advantage of this approach over installing a file-server is that the hard drives are seen directly by the OS, this means you can boot of them, do low level manipulations (like formatting them), directly from the client machine.
One such distribution I would like to see (one can dream), is that changes an old PC into a advanced dock for a laptop. Typically, a thing I can connect my laptop to, and I get to use the peripherals of the PC: hard-drives, CD-drive, keyboard, mouse, heck even the floppy drive.
The goal would be to have all those devices recognised as being devices attached to the laptop, i.e no client server protocol to share them - basically the PC would pretend to be a firewire or USB hub with a lot of devices attached.
Maybe I would even make sense to use the sound-card of the PC. Not so long ago I would have suggested also to access the display of the PC and use its CPU as a powerful GPU, but given the graphical cards one can now find in laptops, this makes no sense.
The question is, how many of your devices have switching power supplies, you desktop computer certainly does, what about your alarm clock, your television, or your cheap printer?
Quite a few years ago, they built a 'ecological house' in Switzerland (sorry I can't seem to be able to find a link). One of the interesting aspects of this house was that it had twin electrical wiring, one 220V AC circuit and one one 12V DC circuit. The 12V circuit was powered by batteries and solar panels, while the 220V circuit was powered by the grid. The point was that converting from 220V to 12 is not very efficient, and solar panels are better suited for producing 12 V DC.
At that time, LED based illumination was not possible. Now if light can be produced efficiently from 12V the list of devices that really need 220V is not that large: mostly cleaning machines and kitchen appliances. Of course rewiring houses is the real problem...
Thanks for the idea. I tried it, problem is I still get some noise, I live near Geneva, and half of the google news are about the Geneva convention, but it is a nice start.
Same thing here, I used to switch between the Swiss, French and US google news sites. Now I can use a single page with the news I am interested in.
One nice feature would be able to select news geographically (in particular for local news). Countries are not really a good granularity when you are close to the border. I suppose the ideal solution would be to have new geo-tagged, but we are not here yet.
Actually I read an optimisation on some darwin mailing list were the altivec register are not immediately save on a context switch, but configured in such a way that a write to those causes an interruption, and that the service routine does the register saving then. The idea was that often, you are prempted by a task that does not use the altivec register, so saving and restoring those registers would be a waste.
I can't remember if this was implemented in the end...
I must agree. Even if a postscript program eventually terminates, it can take ages to render. So a very short file can be very expensive to display.
People who thing this is a purely intellectual issue should try to open this small postscript file. While the file itself is less than 1600 bytes, it should keep your processor busy for a little time (don't worry it terminates).
For those who have a very fast processor, they can simply edit the file (it is a text file) and change the value 14 at line 9 by a larger number. It will take more time but the picture will be nicer.
By the way, this file not only demonstrate that you cannot determine the length of calculation before hand, but also that the resulting image is not fixed, you will get a slightly different image each time you run it.
To what extreme does this go? For example, where is the standard C library?
On Mac OS X the C library is in the System Framework. Frameworks are bundles that contain shared libraries and support files. Frameworks can be either in/System/Library/Frameworks/,/Library/Frameworks/,/Network/Library/Frameworks/ ~/Library/Frameworks/ or inside the application.
Suppose there's a major security flaw in a reasonably popular library. If each package must keep everything inside its own folders, then the library update only goes to apps which are maintained actively -- and which noticed that the library was updated.
You update the System framework.
If, on the other hand, we use traditional UNIX, then one file is replaced in/lib, and at worst we get a warning that something some program is doing with that library is depricated and will be removed. But this gives the individual program maintainers more time to update, because they don't have to rush things out the door to make the security patch. They have until the next library release to get with the program.
The idea of bundles is not so much to replace the Unix way than to improve it. To simplify management and define where things are. The idea of the bundle is that is allows you to store libraries within your application bundle it is in no way an obligation. You can decide to share your library or not and where this shared library is installed (common, network, or user).
And, resource management DOES matter. There is no good reason that my dad, a commodity/stock broker, needs 512 megs of RAM on his machine -- except for the use of this kind of design. It's not just how much memory it takes up on disk, if you have to load glibc fifty times into RAM, you've got problems.
This is true, but shared library are only more economical if they are used by more than one application. This is certainly the case with LibC, but go over the libraries in/usr/local/lib and ask yourself, how many of theose libraries are really used by more than one program?
I have a magic box on my desk where I just lift part of it, type some numbers, and I get connected to the person I want to talk to. There's no IM but it's really easy to use. It doesn't even require a power connection, and it's super-simple. Only a "hangup" button and the numbers. Maybe Apple designed it?
Actually, Apple designed another model, with a single hangup button, the numbers are entered using a rotary dial not unlike the iPod's scroll-wheel.
This model seems to have smaller a smaller market share, with multi-button devices taking over..
Who exactly is behind this FHS initiative? The linked web-site is scant on details and information. There seem to be no know organisation sponsoring this operation and the only other thing on this web site are Enya lyrics (which is nice, I like Enya, but still).
In fact, there is little in OS X that requires 3D support, and OS X uses only a small part of OpenGL. So, a card with good 2D support, antialiasing, and transparency would be just great for desktop use.
I fear that Apple already uses more than what you describe: the screen transitions are full 3D effects and the minimisation effects are done using shaders. I recently went to a talk by Jordan Hubbard and he explained that Apple is planning on offloading more and more processing to the GPU - basically because the processing power was available. If you have a look core image you will see that the features of the graphic card that will be leveraged by Mac OS X 10.4 are well beyond 2D, antialiasing and transparency.
I just had the same thought...
Boy, how computers graphics have changed, and still, the wraith thing in forbidden forest the scariest thing I had in computer games (except maybe the nazi voices in Wolfenstein).
I'm really getting old...
Besides it would make sense to do both the only real difference is the UI programming at least for OS X, the filesystems on both systems are very similar.
Yes and no. Yes, you access both filesystems using the same API (POSIX), but the underlying file-system are quite different. Because of this, if you are going to search / index the filesystem, you will have to do it differently if you want to be efficient (the Mac OS X find utility does not use either the find or locate command-line tools).
Most importantly, this is not about API, this is about data. What this is all about is searching and indexing datafiles and from this point of view the files on a typical Mac OS X machine and a Linux desktop will be quite different.
For instance on Mac OS X, some data files are actually bundles, i.e a directory with a special bit telling the Finder to handle the folder as a single file. Keynotes files are bundles with extension.key that contain an XML manifest an the different files included in the presentation. Older Mac OS filetypes would store some meta-data (icons, keywords) in the resource forks. Those things have, as far as I know, no equivalent in the Linux world.
On the other hand, a Linux version would have to cope with the differences between distributions (what source code should be indexed on gentoo machine?) , the different desktop managers (they might store interesting information), and different file format (it would be nice if it could parse tgif files for instance).
In the end, it is all about data, not about licences, APIs or anything else. The whole point of meta-data and searching, for me, is not about indexing my music collection (I keep it organised), but to be able to search my old files, which include Quickdraw 1 Picts and Word 4.0 (DOS) files.
Of Darwin he speaks as though it were a complete system and not an incomplete husk of one. He even calls it a Unix, while it is not. His views seem tainted and hazed by his own prejudices.
MMh, I think he is not the only one. While I agree that the article is quite bad and not mentioning DragonFly is a clear shortcoming, dismissing Darwin as being an incomplete husk without any arguments sounds quite prejudiced to me...
I see two issues for running NT/PPC in an emulation layer on a Mac.
The type of PPC CPUs that NT actually expect. While the PPC ISA is the same in user mode, it can be quite different in supervisor mode, so the OS might need to be adapted a bit (or the emulator / virtualizer needs to do some trickery).
The hardware drivers. The Windows NT kernel will have drivers for some kind of hardware, this means you need to implement some fake hardware in the virtualisation layer or implement windows NT/PPC drivers for virtualisation environnement.
Assuming you would recompile Win32 code for PPC using this setting, the resulting code would be optimised for old PPC processors and the whole GUI would run via en emulated video device. You might be better of using Wine with OS X. At least you could compile with a recent compiler.
Just my two cents...
Slashdot Mirror
Actually, I'm I'm not fully convinced that the mbox format cannot be indexed by spotlight. The contact database of address book is also in a single file and it can be searched. If I remember Ars Technica's excellent article on Tiger, there is a trick to do this.
I know the source is available, but somehow, I have trouble seeing Linux adopting a launch daemon released under Apple's open source license...
The advantages of this approach are the following:
- All those daemon have shared functionality but different code, this means more code to maintain and more security risks (all those daemon have to run at some point as root). Each of the 'old' daemon also has a different file format, which does not make life easier either.
- Process launch criteria can be a mix of those offered by the 'old' daemons, crond is based on time, inetd on network connections, initd on boot sequence. Launchd 'understands" all those notions.
The example I was given was of starting a backup task on a laptop. The criteria to launch the task would be something like if last execution was more than X days ago and the computer is not running on battery power and there is a network connection and CPU load has been low for some time then launch... Launchd is supposed to have 'adapters' to understand the old file formats.Personally I think this is a good idea, factoring out common functionality and using more reasonable file formats, but of course the old guard will complain that the current set of daemon just works (not on a laptop) and that this was proposed by people who do not understand Unix - who is this Jordan Hubbard anyway? :-)
The privacy problem in Office Documents was mostly that the file contained data that should not have been there. Somehow, if the meta-information extractors can extract more information that the user can by opening the file with the appropriate application, then there is something fishy in your application and is IMHO not a meta data problem per se.
I would be more interested to see if the meta-index respects the permissions (Unix or ACLs), i.e that an user can only see meta-information of files he can read.
I agree, if more articles linked by slashdot were of this quality, I would sure be happy. Then again I'm really disappointed by the discussion here on slashdot, we get a really technical article (stuff for that matters), and people keep bitching about the price of the upgrade, making silly wishes for g5 laptops or OS for intel.
A few months ago, Jordan Hubbard came to CERN to talk about some of the Unix elements of Tiger, and talked about launchd. I think that this is one of the features of Tiger that should be cloned ported to Linux (John Siracusa seems to agree). Having an unified launching mechanism for processes is really something that is needed on Unix, especially for laptops.
You really want to be able to launch processes depending on different triggers and circumstances, like saying at that time, if the machine has been idle for some time and I'm not running on battery power, then launch that process. Yes, you can do hack similar functionality with scripts, but no, this is not convenient or stable.
As people have more and more old machines lying around, having a live CD with a single purpose operating system. I think there are already distributions that transform a PC into a router, but changing an old PC into a firewire hard drive enclosure looks like an interesting idea.
One advantage of this approach over installing a file-server is that the hard drives are seen directly by the OS, this means you can boot of them, do low level manipulations (like formatting them), directly from the client machine.
One such distribution I would like to see (one can dream), is that changes an old PC into a advanced dock for a laptop. Typically, a thing I can connect my laptop to, and I get to use the peripherals of the PC: hard-drives, CD-drive, keyboard, mouse, heck even the floppy drive.
The goal would be to have all those devices recognised as being devices attached to the laptop, i.e no client server protocol to share them - basically the PC would pretend to be a firewire or USB hub with a lot of devices attached.
Maybe I would even make sense to use the sound-card of the PC. Not so long ago I would have suggested also to access the display of the PC and use its CPU as a powerful GPU, but given the graphical cards one can now find in laptops, this makes no sense.
The question is, how many of your devices have switching power supplies, you desktop computer certainly does, what about your alarm clock, your television, or your cheap printer?
Quite a few years ago, they built a 'ecological house' in Switzerland (sorry I can't seem to be able to find a link). One of the interesting aspects of this house was that it had twin electrical wiring, one 220V AC circuit and one one 12V DC circuit. The 12V circuit was powered by batteries and solar panels, while the 220V circuit was powered by the grid. The point was that converting from 220V to 12 is not very efficient, and solar panels are better suited for producing 12 V DC.
At that time, LED based illumination was not possible. Now if light can be produced efficiently from 12V the list of devices that really need 220V is not that large: mostly cleaning machines and kitchen appliances. Of course rewiring houses is the real problem...
Information about the changes in Safari 1.3 can be found on on David Hyatt's blog.
Thanks for the idea. I tried it, problem is I still get some noise, I live near Geneva, and half of the google news are about the Geneva convention, but it is a nice start.
One nice feature would be able to select news geographically (in particular for local news). Countries are not really a good granularity when you are close to the border. I suppose the ideal solution would be to have new geo-tagged, but we are not here yet.
I can't remember if this was implemented in the end...
People who thing this is a purely intellectual issue should try to open this small postscript file. While the file itself is less than 1600 bytes, it should keep your processor busy for a little time (don't worry it terminates).
For those who have a very fast processor, they can simply edit the file (it is a text file) and change the value 14 at line 9 by a larger number. It will take more time but the picture will be nicer.
By the way, this file not only demonstrate that you cannot determine the length of calculation before hand, but also that the resulting image is not fixed, you will get a slightly different image each time you run it.
Thanks for the answer. I suppose this explains why OS X does not comply with this standard...
Who exactly is behind this FHS initiative? The linked web-site is scant on details and information. There seem to be no know organisation sponsoring this operation and the only other thing on this web site are Enya lyrics (which is nice, I like Enya, but still).
I just had the same thought...
Boy, how computers graphics have changed, and still, the wraith thing in forbidden forest the scariest thing I had in computer games (except maybe the nazi voices in Wolfenstein).
I'm really getting old...
A drawing tool.
Most importantly, this is not about API, this is about data. What this is all about is searching and indexing datafiles and from this point of view the files on a typical Mac OS X machine and a Linux desktop will be quite different.
For instance on Mac OS X, some data files are actually bundles, i.e a directory with a special bit telling the Finder to handle the folder as a single file. Keynotes files are bundles with extension .key that contain an XML manifest an the different files included in the presentation. Older Mac OS filetypes would store some meta-data (icons, keywords) in the resource forks. Those things have, as far as I know, no equivalent in the Linux world.
On the other hand, a Linux version would have to cope with the differences between distributions (what source code should be indexed on gentoo machine?) , the different desktop managers (they might store interesting information), and different file format (it would be nice if it could parse tgif files for instance).
In the end, it is all about data, not about licences, APIs or anything else. The whole point of meta-data and searching, for me, is not about indexing my music collection (I keep it organised), but to be able to search my old files, which include Quickdraw 1 Picts and Word 4.0 (DOS) files.
Things might change: http://darwine.opendarwin.org/
- The type of PPC CPUs that NT actually expect. While the PPC ISA is the same in user mode, it can be quite different in supervisor mode, so the OS might need to be adapted a bit (or the emulator / virtualizer needs to do some trickery).
- The hardware drivers. The Windows NT kernel will have drivers for some kind of hardware, this means you need to implement some fake hardware in the virtualisation layer or implement windows NT/PPC drivers for virtualisation environnement.
Assuming you would recompile Win32 code for PPC using this setting, the resulting code would be optimised for old PPC processors and the whole GUI would run via en emulated video device. You might be better of using Wine with OS X. At least you could compile with a recent compiler.Just my two cents...