Then there's no reason to mention how "lean" the compiler is. Yes you can optimize functions in the codebase, you can also optimize the compiler to recognize and convert the codebase to more efficient machine code. Two paths to the same end.
The PowerPC compilers for GCC have been lagging behind in performance behind the Intel versions for some time now. That is a known commodity. Optimization of code or not, I would suffice to say the x86 platform has gotten much more focus considering, from a hardware standpoint, it has had more volume sales. Whether GCC has improved since Apple left PPC remains to be seen... one could argue that with the Wii and X360 both being PPC machines, they might've led to GCC on PPC improvements as the volume sales of both are considerable (esp.compared to the whole Mac market as a whole). Then again, both platforms have their own unique ASIC's so how much that would play into Apple's hands is still questionable.
Not entirely true. They are also focusing on stability which could be completely architecture agnostic, bad functions, memory leaks, etc. Fixing those as they are identified could lead to a much more stable, abeit not much faster, PPC release.
There is nothing to say that, going forward, a similar emphasis is being placed on Leopard as well simultaneously to Snow Leopard. After all, Apple did just release a colossal upgrade to 10.5 in the form of 10.5.3 that fixed many bugs. The fact that 10.5.4 was just seeded still brings hope that any bugs for Leopard will receive a similar attention focus as Snow Leopard will be on optimizing performance for Intel. I, admittedly, am hoping that there's a significant iTunes upgrade coming soon that rectifies some of the issues I and others are having with it. Then again... I can't wait for a Cocoa iTunes which I feel is bound to happen eventually.
What other significant indications are there other than the developer preview requiring an Intel machine? Yes they are focusing most of their optimization on Intel and some of the biggest gains/technologies used there will only work on Intel and that I can understand. That does not necessarily mean that 10.6 will be Intel only.
Umm... there is little rationale to put out a developer preview that is Intel-only if the final end result isn't to be Intel-only. I mean, think about that for a second. If they wanted developers to continue supporting PPC in 10.6 going forward, you would definitely want your 10.6 build to be capable of running on PowerPC as well so you can compile and test said build on a PowerPC installed version of 10.6...
That all depends on how much performance you wanted to tweak out of your application. For many small applications, yes a simple check box will do it. For some others it's not that simple. And every universal binary generated will take time to test and debug on a PPC machine. You don't compile code for multiple system architectures and only test one hoping that the other works. There is plenty of time and money to be saved by going Intel only and the fact that no PPC support was not announced officially is a big indication that there's a good chance it will be there.
That is true regardless though. Most developers do not just build applications that are "Leopard-only" in as much as they try to support as many versions of the platform as they can (within reason), whether that's an after-the-fact maintenance release to an old build or a simultaneous launch of their app. via multiple versions of Apple's tools (and/or Codewarrior). There is nothing to say that an application written has to be specific to Leopard, Snow Leopard, or otherwise. You can create a Universal Binary that supports Snow-Leopard related features but also is savvy enough to run on Leopard, Tiger, Panther, etc. in as much as you want to put the effort and time into dedicating for said machines. That is always the developers' call...
And for developers who want code that is Universal they now have to be sure that their application runs on multiple OS versions, where if PPC is supported in 10.6 would allow universals to spec at 10.6 or later to avoid possible issues that might arise trying to use the more optimized API environment that was not available on 10.5.
Umm... they've had to do that if they wanted to support 10.2, 10.3, 10.4, 10.5, etc. What's the difference really? I don't necessarily expect a more optimized API environment in that I suspect effort is being done to make the API's work with the new technologies. In that vain, writing an application for 10.6 may be absolutely little to no difference from Leopard, Tiger, etc. themselves other than to do some calls to work with Grand Central and/or OpenCL where necessary. Then again, there's nothing to say that these technologies aren't done in a way to take advantage of existing code written strictly to the API's themselves and not written to the metal, so to speak. We just don't know that well yet.
But if they're significantly changing the API then it's going to really require a lot of testing to prove that the PPC code still works as intended, either that or you really are going to have to trust your compiler and be ready for a customer support nightmare if/when PPC apps have issues.
Who said they are? You act as if matticus is jumping to presumptions and yet you, yourself, seem inclined to generate presumptions as well. There is no documentation that the API is radically changing. There is no documented proof that the reason Snow Leopard was seeded to developers was anything more than to show them what their apps. will be able to leverage in the future. That is not documented in any way that there is radical rewrites being involved. I wouldn't be surprised, judging by Apple's emphasis on ease of deployment of their tech. shown at WWDC, that this was more done as a platform to test applications to make sure that they don't need clean up for various workarounds developers have pulled out of thin air themselves to make their apps. work. Rather than using some form of plug-in or shim for their apps., as they may have in the past, and which may/may not cause conflicts in the new OS... it is possible to believe that going forward with Snow Leopard, they might need to remove those calls to leverage Apple's in-built technologies and prevent conflicts with them. This wouldn't "magically" change the API's unless Apple leveraged any new API's to do so and there's nothing that says that Apple hasn't proofed the tech so that the deployment of said application can't be savvy enough to tell whether or not the tech exists and/or treat it as a single core unit.
After all, the release says "Intel-only" but does not stipulate if it's Intel multicore or not. Core Solo Macs do exist and if they're supported... well... odds are OpenCL and Grand Central will be of little consequence to them. You would want to be compatible with them if the OS is, correct? If OpenCL and/or Grand Central aren't capable on said machines or have no purpose... then the software obviously has to be savvy enough to know when to say "There isn't any cores to use" and allow the app. to still operate and work. If it can do that... why can't the apps. just run out of 10.5 or 10.4 then if the core API's didn't change?
If disk space and RAM can change why not the ability to run on PPC? Maybe Apple doesn't know for sure yet if it's going to be included or not.
Because much of the new tech being put in has no bearing on PPC. Why write access to old hardware with a new OS if you're trying to clean house and futureproof yourself? If Snow Leopard buys little over Leopard, if the API's aren't changing radically and if you can still build Universal binaries (that likely will leverage PPC compiling and perhaps even be savvy enough to build to the last known PPC version of the API's)... there's no real reason for a PPC version of Snow Leopard. Esp. since the debugging for Leopard has already been put into full speed ahead mode judging by 10.5.3.
I do realize that but just because the current release is Intel only everyone is taking it to be gospel that PPC support is dead. Can you at least admit that because of no official announcement and the fact that this is a very early release that PPC is not necessarily out of the question?
Who says PPC support is dead? PPC support for Snow Leopard may be, but that doesn't mean that Apple and others won't continue to support Leopard, Tiger, Panther, et al. going forward on PPC and/or Intel. Nothing says that Apple needs to build their OS'es to support older hardware ad-infinitum while still supporting said platform with software going forward. Moral of the story, there's little logic in Apple putting a dev release out that is Intel-only if the final release weren't going to be Intel-only as well.
If PPC is the source of speed and stability issues as so many seem to think, removing Rosetta will give users a better impression of speed and improve stability since apps will all have to run natively which reduces the risk of Rosetta translation being a stability issue. As everyone with PPC has been told here, it's not like 10.5 with Rosetta will stop working.
Suicidal... not every application has been ported completely yet. Not everyone has bought Intel-only software yet. This is like Classic in a sense, Classic was supported for a lonnnnnnng time. Rosetta will likely be there for quite awhile as well while people make the transition as their budget allows. I expect Snow Leopard to have it, and I expect the performance efforts done to streamline Snow Leopard might even make Rosetta run even better going forward. That said, there are situations where a Rosetta-operated app. will not work well and that is documented. Anything calling AltiVec/SIMD for PowerPC is generally going to have problems and likely have stability issues.
All of that said... my impression of Leopard is unlike others. I find it very stable on all of the machines it is running on. I do have a few quirks, but they're more with applications acting odd moreso than the OS itself which is pretty fast and stable for me on the 3 machines I have it on. My 24" iMac came with it and runs it beautifully. The Intel iMac that came with it runs it beautifully. The G4 iMac I have it on runs Leopard faster than Tiger ever ran on it, and I did manage to do a fresh install vs. an archive and install to know whether it's fast or not.
That said, I do have a weird iTunes issue where the disc mounts, CDDB tries to access it's DB and iTunes unmounts the disc. Ejecting and reinserting the disc, the disc mounts and CDDB finds the info. and I can then rip the disc to the library. It is a weird annoyance that only has happened in Leopard to date, but once again... it is also one of the few mainstream/free Apps. put out by Apple that to this date has remained a Carbon app. All of iLife is Cocoa with exception to it. Safari, is also Cocoa. Safari, despite others' claims to the contrary, is very stable for me.
Why would I want to spend my time working on developing applications for 10.6 and testing the PPC side of a universal binary until the beta comes out? I'd want to know ASAP if the PPC were being dropped and the fact that it hasn't been announced makes the possibility of Snow Leopard supporting it remains. Otherwise there is no point in developing a 10.6 only application that has to be a universal binary.
Because nothing has been said that the API's are so radically different that an app. written and compiled in 10.6's dev tools won't work in 10.5. In contrast, as I would suggest... the general idea is to make sure that 10.4 and 10.5-based codebases work with 10.6 in the native environment it is going to be available in. MP, thus far, has been largely enabled by the developer making their applications threaded and not necessarily using Apple's tools to make that happen. With 10.6 and Grand Central and OpenCL... those that have written apps. to be multi-threaded and work with previous OS builds might now face major changes going forward for 10.6 and beyond. The general idea is that they're future proofing the OS by helping developers remain compatible ahead of shipment. I'm sure that the developers will know soon enough if a compiled app. in 10.6 works well in 10.5, 10.4, etc. as most likely have both dev environments at their disposal... esp. now. It's not like those machines they brought their shiny 10.6 development discs home to weren't running some other version of Mac OS prior to them leaving for WWDC...
You're convinced it's gone, I think it will be there and it seems obvious that neither of us will be convinced one way or the other until the official announcement is done.
I'm convinced it's gone because there's little logic in a dev platform being Intel-only if the eventual plan is for the final release to be PPC and Intel. LoL It's pretty simple to deduce, esp. since the rumors have pointed in that direction long since prior to Snow Leopard being acknowledged by Apple and the documentation put-out documenting that... ummm... it's Intel-only. The rumors that called for Snow Leopard (by name) as 10.6 all said that it was Intel-only and the dev release magically is Snow Leopard (same name as the rumors) and it's requirements say "Intel-only." That's enough to seal the deal for me on it. It also makes quite a bit of sense considering that no *MAJOR* new end-user features are going to be there other than Quicktime X and the other features involved are only likely to be leveraged by modern (Intel) hardware.
I didn't think you were both feuding... just having an educated and mature discussion. I don't see it as a war of any sort. I do tend to agree more with matticus because I feel his logic makes much more sense. That said, I don't think that makes you a bad guy. LoL The beauty of a messageboard is when you can have a spirited debate and still shake hands at the end though.
