Panther wasn't the best OS ever. I don't recall the specific issues, but while 10.2 was a big step over 10.1, I don't recall really being satisfied with an OS after that until 10.4. I wish it were easier to recall specifics....
-and-
I did get the impression that 10.3 was a bit of a lemon.
I've had the opportunity to work with most of the operating systems in the developmental history of Mac OS X (including many which were never released to the general public). The one trait that all of these operating systems have in common is stability. If run on the proper hardware, all of these operating systems have been very stable. Now just saying that I've used them isn't really enough to drive the point home here, so here is a list of the ones I have had extensive experience using:
- NEXTSTEP 2.0
- NEXTSTEP 2.1
- NEXTSTEP 3.1
- NEXTSTEP 3.2
- NEXTSTEP 3.3
- OPENSTEP 4.1
- OPENSTEP 4.2
- Rhapsody 5.0 (Rhapsody Developer Release)
- Rhapsody 5.1 (Rhapsody Developer Release 2)
- Rhapsody 5.3-5.5 (Mac OS X Server 1.0.x)
- Rhapsody 5.6 (Mac OS X Server 1.2 & 1.2v3)
- Mac OS X Developer Preview 3
- Mac OS X Developer Preview 4
- Mac OS X Public Beta
- Mac OS X v10.0.x
- Mac OS X v10.1.x
- Mac OS X v10.2.x
- Mac OS X v10.3.x
- Mac OS X v10.4.x
- Mac OS X v10.5.x
Now, in addition to that the ones in bold are ones that I still use on my systems today. So in many cases we aren't talking about dusty memories of the past, they are still asked to pull their own weight (some more than a decade beyond when most other users had moved on).
Were there ones that I wouldn't use to aid me in getting work done? Absolutely, which is why there are gaps between some of the releases I still currently use on that list. But none of those gaps are due to the base OS, but rather changes happening to the application environment. An operating system is, after all, only as good as the applications which can be run on it. And the best operating system in the world without applications is really pretty useless.
Some of the transitions were harder (and longer) than others. The move from NEXTSTEP to OPENSTEP was relatively painless, but that was because both Sun and NeXT spent a good deal of time making sure to not damage what worked while creating the
OpenStep Specifications, and then not removing backwards compatibility while applying those specifications to OPENSTEP 4.x.
The move from OPENSTEP to Rhapsody was more painful because not only was backwards compatibility lost, the foundations for applications were expanded and the interface was changed. So while the first version of Rhapsody may have been rock solid, it was also nearly as useful as a rock. While missing some of the refinements of later versions, the next version of Rhapsody was a very usable operating system (which I have been using consistently on an IBM ThinkPad since 1999 for actual work), and this was mainly because it was a good environment for running good applications (which I happen to have quite a few). By the final release of Rhapsody (then called Mac OS X Server 1.2v3) in 2000, the environment had become a great place to do work in and developers had put together some great applications for it. I have two systems (a Power Macintosh 8600/300 and a PowerBook G3/300) that are still running this operating system today.
But neither of those transitions faced the daunting task of moving from Rhapsody to Mac OS X. I had (and was giving Apple feedback on) many of the early versions of Mac OS X... but I sure wasn't using them to get any work done (my Rhapsody and Mac OS 8/9 systems were where I was doing my actual computing). That transition had some massive hurdles to overcome.
First of these was removing all the 4.4BSD (encumbered) elements and replacing them with elements from 4.4BSD Lite, FreeBSD, NetBSD and OpenBSD. This created Darwin and meant that Apple would no longer have to pay the Regents of the University of California for every copy of the operating system sold (based on the license agreement that NeXT had entered into in the late 1980s). That part was actually relatively painless and didn't have a major impact on the user environment (the first Mac OS X Developer Preview seemed almost identical to the concurrently shipping Rhapsody based Mac OS X Server 1.0.x).
The next wasn't so easy. Back in the days of Apple's Copland project, developers told Apple in no uncertain terms that they would
not do extensive rewrites of their applications for a new Apple operating system. It didn't matter if those applications would benefit from the new operating system, they couldn't justify the time and effort. When the developers (like Adobe, Microsoft and Macromedia) took the same stand with Rhapsody, it caught a lot of people at Apple (who had come from NeXT) by surprise. The solution was to use what had been started in the Copland project in Rhapsody, and this was later called Carbon. Within the span of a few weeks Apple graft a version of Carbon into Rhapsody 5.1, they ported AppleWorks 5 and the lead Photoshop developer at Adobe ported Photoshop 5.0 and displayed those apps running as
native applications in Rhapsody at WWDC 98. Even with those successful examples, Carbon wasn't ready. So Apple started developers off by including Carbon support in Mac OS 8.5 and continued to refine the Mac OS X version.
What a lot of people don't generally know is that developers were afraid that Carbon applications were going to be second class citizens in the new operating system. To prove it's commitment to Carbon, Apple made the most important application on the Mac in Carbon... the Finder. This wasn't a port of the Finder from Mac OS 8/9, this was a ground up new application which actually help refine Carbon. But at the same time made Mac OS X feel slower than previous versions in those early days.
At about the same time Apple realized that they could save more money by removing more of the licensed elements from the old NeXT days. NEXTSTEP, OPENSTEP and Rhapsody all used Display Postscript. Display Postscript was codeveloped by NeXT and Adobe, but NeXT and Apple still had to pay a license fee for every copy of the operating system sold. Apple replaced it with Display PDF (part of Apple's Quartz graphics). Between not having the level of refinement that Display Postscript had and the Mac OS X interface developers putting transparencies all over the place, this added another level of burden to Mac OS X.
Because most of the developers who made the applications I used in Rhapsody were making versions of these same applications for Mac OS X, I could use it early on to a reasonable degree... but there wasn't a good reason to stop using (or move any of my activities) from Rhapsody to Mac OS X back then. But what I had been hoping for was more of a replacement for Mac OS 8/9. My main Mac OS 8/9 applications run great in Rhapsody's Blue Box environment (and I still do that with many today), but having many of these apps running native together would have been great (and what I was wanting from Mac OS X). High on that list of apps was Photoshop.
As I pointed out earlier, one person at Adobe had ported Photoshop 5.0 to Carbon in Rhapsody within a few weeks back in 1998. One would think that Photoshop would have been the first Mac OS X native application by Adobe. It wasn't. Adobe ported Acrobat, Illustrator, InDesign, Premiere, AfterEffects, GoLive and LiveMotion to Mac OS X
before Photoshop. Photoshop for Mac OS X was released at about the same time as Mac OS X v10.2. It seems that Adobe wasn't very happy about Apple replacing Display Postscript in Mac OS X (which would have been a nice source of revenue for Adobe), and holding back Photoshop was Adobe's way of punishing Apple.
So Mac OS X v10.2 was the first version of Mac OS X that I used for regular productive computing, and I still make use of it today (and it feels less cluttered than later versions to me). And the only reason I use 10.3 and 10.4 (and eventually 10.5) is application compatibility. The operating system itself (which I feel should do it's job well and otherwise stay out of the way) was pretty much finished for me in 10.2.
This isn't to say that there aren't great advances in later versions. One of the major philosophies from NeXT was that applications shouldn't have to
reinvent the wheel for many elements. Why (for example) should each application have it's own spellchecking engine and database? Why not have one that all applications can use. And if one application does something well, why not share that ability with other applications. This is the concept behind services (I've written about this in the NeXT environment
here). Apple has expanded on Services 10 fold in 10.5, and more and more developers are starting to take advantage of them. Sadly Services haven't been largely useable in Carbon applications, but 10.5 is making a turning point in that.
But back to the point at hand, I didn't have any issues with stability in even the beta versions of Mac OS X I was given to critique. Usually it was other factors at play that left me dissatisfied in one way or another.
Now, can someone FUBAR a Mac OS X system? Absolutely! And Macs that are open to many users under the administration of inexperienced admins are perfect examples of this... But then again, Windows systems under the same conditions are just as bad (which is why you'll never see me using the
disingenuous tactic of holding a Windows system in a computer lab up as an example of their general stability).
As for 10.3 specifically... it is still being used by a majority of my clients (and I have it on three of my systems) without any issues what so ever. The only problems I've seen came from people attempting to run applications designed for 10.4 in 10.3... but I saw the same issues with people attempting to run apps designed for 10.3 in 10.2. Otherwise those have been absolutely rock solid releases. My clients encountered more issues with 10.5.0/10.5.1 than they had with the releases of 10.2, 10.3 and 10.4 combined. But 10.5 was a pretty big change.
All
lemons should be this bad...