Speculation about Apple... and Sun
By ColmSmyth on Jan 04, 2005
MacNewsWorld hosts an article that is always interesting, sometimes insightful and occasionally a little wrong in my view, but well worth reading just the same. Paul Murphy offers a broad analysis that spans hardware and software. Here's my play-by-play feedback and 2 cents...
- Paul's comments on Sun's throughput computing capability are right on; but don't forget Sun's x86 capability with Solaris x86 for high performance on low-end servers and (most definitely along with Linux) especially for desktops (via Java Desktop System); also, SunRay runs JDS extremely well - startup performance is even better than a standalone desktop (because application binaries are almost always already in memory) and the shared SunRay CPUs give you extra desktop power when you need it, not to mention better security, reliability and the downright damn beautiful experience of connecting to your desktop by just putting a card into any SunRay desktop in an office or on a campus - I never get tired of that, nor the delighted amazement when a new employee at our office experiences it for the first time
- Paul speculates about IBM's evolution path for PowerPC which Apple depends on; I think porting the Mac OS shell is not the key issue (any properly layered software stack is pretty portable), but maintaining 100% compatibility with the Mac OS platform API's and the porting impact on ISV's developing applications for Mac OS might be more of an issue; I note that IBM do have a desktop/portal hyrbrid product (Workplace, based incidentally on (an old version of) OpenOffice.org) so Apple's boxes don't only compete in the server space
- He also raises the old chestnut of Mac OS desktop running on Sun hardware, especially SunRay; Mac OS certainly has cool coming out of its pixels, but don't overlook the importance of a user experience that facilitates migration from Windows (JDS has this in spades)
- He also speculates on the value of a Mac OS graphics porting layer to allow say GNOME-based apps to run on it; this certainly makes sense as it gives ISVs a broader platform to target and the Gtk-Win32 port shows that this is not only possible but practical (though it would need more vendor commitment which Gtk-Win32 does not yet have enough of)
I hope you've enjoyed this cocktail of thoughts and speculations from Paul Murphy and I - but what do you think?