NeXT Step, OpenStep and Mac OS X can, in fact, do what you are describing (using "bundles"). However, Darwin (by itself) doesn't do this. The "mach-o" binary format (supported by all the operating systems you mentioned) can store "fat-binaries" in a single executable file, though. Apple also supports writing your "Cocoa" applications in Java.
Since Mac OS X only supports Apple branded PPC hardware currently, these features are somewhat wasted. I personally would love to see Apple support other types of hardware (SPARC, MIPS, Alpha, PA-RISC, IA-64, and even x86), but I'm not holding my breath. It appears that (for now) Apple is using Mac OS X to sell PPC hardware (which I like much better than x86). Until that changes, I wouldn't expect to see support for competing hardware (IA-64 or x86). Porting to any other architectures (i.e. SPARC) probably wouldn't generate enough cash to justify the effort.
I do believe that it would be possible to run Mac OS X on other PPC machines however. Since Darwin is open sourced, the kernel could be ported to say an AS/400 or a PReP/PPCP workstation and since Mac OS X is binary compatible, it would probably be possible to run it.
As for non-PPC hardware, I've been thinking that if you managed to get Darwin ported to SPARC (or any other instruction-set), it might be possible to combine it with an emulator or a "Dynamic Cross-Processor Binary Translator" to be able to run the rest of Mac OS X and other PPC software written for it (i.e. Photoshop). It may be even possible to store that translated code to "fat-binary" mach-o files for native execution at a later date.
Once the whole environment was working, gcc could be used to compile applications (when source is available or developers willing) to natively support that hardware. Since Mac OS X uses dynamic linking, it may even be possible to mix and match native and emulated code in a single application (PPC application, SPARC pluggins).
Something like this would take an enormous amount of effort though. It's probably a lot easier to just go out and buy the latest and greatest from Apple, which is pretty competitive with the other workstation hardware in it's price range.
In Hamden Connecticut, USA, SNET (Southern New England Telephone) is the xDSL provider and default ISP (for xDSL). You can get ADSL with SNET as the ISP for $40/month. This is about the same cost as having a second phone line (~$20/month) and a dialup ISP (~$20/month).
The actual ADSL connection is up to 6 Mbps downstream and 384 kbps upstream depending on line conditions. However, at the $40 price range, SNET caps your bandwidth to 1.5 Mbps downstream and 128 kbps upstream. They allow multiple simultaneous PPPoE streams (over the same or different DSL connection) and the service includes a regular dialup account with 5 email boxes.
SNET also offers two other pricing plans for xDSL (that I'm aware of). For $80/month you get exactly the same services as the $40 plan with the addition of a static IP address (I'm not sure about the multiple PPPoE streams with this account). They also offer a higher bandwidth option for, I believe, $120/month. This removes the bandwidth caps from your connection (allowing up to 6 Mbps down/384 kbps up) and gives you a static IP. As far as I know, SNET does not currently offer SDSL.
I've had the $80/month plan for just over a year, and I have been very pleased by the service. My phone line is clean and relatively short, so it supports the full 6 Mbps downstream and 384 kbps upstream. Of coarse I'm capped, so I don't see any more than 1.5 Mbps down (183 kBytes/sec) and 128 kbps up (15.6 kBytes/sec), but I usually don't see much less either.
I usually get about 150 kBytes/sec downloads (www, ftp, etc.) from most sites (sometimes more, sometimes less). I'd be willing to bet that it's not my connection slowing me down in most cases. SNET is now owned by SBC Communications which I believe is based in Texas (and actually provides the DSL service). Often, when I do a traceroute, and I'm trying to connect to a machine that is relatively local, the route goes down to New York, then Texas, bounces around the mid-west a bit and finally starts heading back towards CT. Even with the poor route, the connection seems to be OK.
The two things I'm not happy about are PPPoE and the 128kbps upstream cap. I've talked with 3rd level SNET xDSL support and they confirmed my suspicion that there is no technological reason to use PPPoE. Apparently, the reason is that SBC Communications "says so". Well, at least, their upper management wants certain types of reporting which was only achievable by using PPPoE. This is supposedly changing and maybe someday they will offer strait IP over DSL with DHCP for IP address assignment and authentication (the way it ought to be).
I experienced at least one problem that wouldn't have effected me if it weren't for PPPoE. My friends with dynamic IPs report all kinds of trouble with PPPoE.
As for my upstream woes, 128kbps is just not enough to serve anything to anyone except a single modem user, with out it being frustratingly slow. I could go for the high priced option, except 384 kbps isn't enough either, so why should I pay 50% more for something that isn't really any better (at least for upstream).
I need SDSL! 1 Mbps downstream and 1 Mbps upstream would be enough for me. I could live with a little less downstream bandwidth if I only had more upstream.
I've always believed this. I thought this was obvious. Look at the binary number system (or any number system). Each digit doubles the level of complexity. An 8-bit binary number represents 256 possibilities, not 8.
I think this is one of the fundamental flaws of genetic engineering right now. A bunch of (vary knowledgeable) scientists and/or doctors (who think they now more than they actually do) start mucking with genetics because they've noticed a pattern between the presence or absence of a given gene and some property of the life form they are studying.
Slashdot Mirror
Since Mac OS X only supports Apple branded PPC hardware currently, these features are somewhat wasted. I personally would love to see Apple support other types of hardware (SPARC, MIPS, Alpha, PA-RISC, IA-64, and even x86), but I'm not holding my breath. It appears that (for now) Apple is using Mac OS X to sell PPC hardware (which I like much better than x86). Until that changes, I wouldn't expect to see support for competing hardware (IA-64 or x86). Porting to any other architectures (i.e. SPARC) probably wouldn't generate enough cash to justify the effort.
I do believe that it would be possible to run Mac OS X on other PPC machines however. Since Darwin is open sourced, the kernel could be ported to say an AS/400 or a PReP/PPCP workstation and since Mac OS X is binary compatible, it would probably be possible to run it.
As for non-PPC hardware, I've been thinking that if you managed to get Darwin ported to SPARC (or any other instruction-set), it might be possible to combine it with an emulator or a "Dynamic Cross-Processor Binary Translator" to be able to run the rest of Mac OS X and other PPC software written for it (i.e. Photoshop). It may be even possible to store that translated code to "fat-binary" mach-o files for native execution at a later date.
Once the whole environment was working, gcc could be used to compile applications (when source is available or developers willing) to natively support that hardware. Since Mac OS X uses dynamic linking, it may even be possible to mix and match native and emulated code in a single application (PPC application, SPARC pluggins).
Something like this would take an enormous amount of effort though. It's probably a lot easier to just go out and buy the latest and greatest from Apple, which is pretty competitive with the other workstation hardware in it's price range.
The actual ADSL connection is up to 6 Mbps downstream and 384 kbps upstream depending on line conditions. However, at the $40 price range, SNET caps your bandwidth to 1.5 Mbps downstream and 128 kbps upstream. They allow multiple simultaneous PPPoE streams (over the same or different DSL connection) and the service includes a regular dialup account with 5 email boxes.
SNET also offers two other pricing plans for xDSL (that I'm aware of). For $80/month you get exactly the same services as the $40 plan with the addition of a static IP address (I'm not sure about the multiple PPPoE streams with this account). They also offer a higher bandwidth option for, I believe, $120/month. This removes the bandwidth caps from your connection (allowing up to 6 Mbps down/384 kbps up) and gives you a static IP. As far as I know, SNET does not currently offer SDSL.
I've had the $80/month plan for just over a year, and I have been very pleased by the service. My phone line is clean and relatively short, so it supports the full 6 Mbps downstream and 384 kbps upstream. Of coarse I'm capped, so I don't see any more than 1.5 Mbps down (183 kBytes/sec) and 128 kbps up (15.6 kBytes/sec), but I usually don't see much less either.
I usually get about 150 kBytes/sec downloads (www, ftp, etc.) from most sites (sometimes more, sometimes less). I'd be willing to bet that it's not my connection slowing me down in most cases. SNET is now owned by SBC Communications which I believe is based in Texas (and actually provides the DSL service). Often, when I do a traceroute, and I'm trying to connect to a machine that is relatively local, the route goes down to New York, then Texas, bounces around the mid-west a bit and finally starts heading back towards CT. Even with the poor route, the connection seems to be OK.
The two things I'm not happy about are PPPoE and the 128kbps upstream cap. I've talked with 3rd level SNET xDSL support and they confirmed my suspicion that there is no technological reason to use PPPoE. Apparently, the reason is that SBC Communications "says so". Well, at least, their upper management wants certain types of reporting which was only achievable by using PPPoE. This is supposedly changing and maybe someday they will offer strait IP over DSL with DHCP for IP address assignment and authentication (the way it ought to be).
I experienced at least one problem that wouldn't have effected me if it weren't for PPPoE. My friends with dynamic IPs report all kinds of trouble with PPPoE.
As for my upstream woes, 128kbps is just not enough to serve anything to anyone except a single modem user, with out it being frustratingly slow. I could go for the high priced option, except 384 kbps isn't enough either, so why should I pay 50% more for something that isn't really any better (at least for upstream).
I need SDSL! 1 Mbps downstream and 1 Mbps upstream would be enough for me. I could live with a little less downstream bandwidth if I only had more upstream.
I think this is one of the fundamental flaws of genetic engineering right now. A bunch of (vary knowledgeable) scientists and/or doctors (who think they now more than they actually do) start mucking with genetics because they've noticed a pattern between the presence or absence of a given gene and some property of the life form they are studying.