But yes, programming for Symbian was quite a steep learning curve. 'Train wreck' might be a bit extreme though, there was method behind the madness.
Basically a total failure of vision regarding how to make the most of the hardware for the benefit of the user.
I don’t disagree with you, but the problem here wasn’t just the software not taking full advantage of the hardware. It was a more fundamental disconnect between hardware and software design. At Nokia, there was practically no innovation where these two were advanced together. Everything was incremental on one side or the other.
But that was never going to last. Once CPU's developed sufficient horsepower and there was enough RAM, the fully fledged OS's underneath Android and iPhone meant they could easily transition to what are now. I doubt Symbian ever could have done that.
I heard devs ranting, but Symbian was very popular, well liked and ahead of anything on the market until it got (almost deservedly) destroyed by Nokia's hubris, Apple and Android.
100% this.
Remember there are 1000x to 10000x as many users as developers. (Extremely conservative estimate.) If users love it that's worth some developer discomfort IMHO.
Active Objects were PITA, but I also admired their efficiency. It was a nice way how to get pseudo-threads on a single core processor. Symbian in general was very efficient with resources. You were able to run fairly complicated algorithms on not-even-200 MHz processor of E52.
From my point of view, there were three major problems with Symbian.
First of all, the dialect of C++ was too distant from normal C++, thus the learning curve for a normal C++ programmer used to standard libraries was enormously steep. Even the GCC that was supplied within the toolchain was always already obsolete, way behind the current GCC version.
Second, the debugging tools were a nightmare. Even on an emulator, you often had to fish for running processes in Windows Task Manager and kill some instances of DE.exe, because otherwise, after the first crash of your app, the debugger would no longer work. On-device debugging was even worse. I think I only saw it working twice, in several years. After some frustration, you just stopped trying to get that damn thing working. And a bad installation of the Nokia USB driver could destabilize your Windows. Compared to that, on-device debugging on Android is just marvellous.
Third, backward compatibility and UI compatibility just wasn't. Nokia constantly churned out new devices by the boatload and all of those had different APIs, screen dimensions, hardware etc. It seemed that their intent was to negate the "write once, run anywhere" Java principle, you had to adjust your apps constantly and at a great expense, otherwise the UX would be horrible, and the Nokia smartphones weren't exactly cheap. Emulators helped maybe with half of that, the rest had to be done on a real device. For a software house, that meant pointless spending on devices that would be obsolete in a year or two, because you couldn't predict in advance which one would be successful and which one would flop.
In all, Symbian would have been fine if someone actually turned it into a developer- and user-friendly operating system. Nokia wasn't really able to do that, even though the foundation base of the OS was pretty solid.
Do you think that would be fixable nowadays? The OS is FOSS now.
Could as much of the weird/obsolete/proprietary stuff as possible be ripped and replaced with current, modern, standardised FOSS stuff, while keeping the tiny elegant OS? Given that backwards compatibility isn't an issue any more, as the existing apps are old and proprietary and irrelevant now?
My personal favourite was my E90 Communicator.
https://en.wikipedia.org/wiki/Nokia_E90_Communicator
In terms of hardware form factor was the best smartphone I have ever owed.
What was good about it:
– Metal case, metal hinge, rock-solid clamshell design
– But small enough to use closed as a candybar phone with T9 text input;
– Good 1-handed and two-handed operation;
– Able to view A4 width PDFs at readable size on internal screen;
– Fast accurate thumb typing on internal QWERTY keyboard;
– Data port plus charging port plus headphone socket.
It wasn't perfect. The external numeric keypad needed better spacing, like a 6310i keypad. The internal QWERTY one the same. It needed USB instead of a Nokia port, and a standard-sized headphone jack. It needed an internal touchscreen. It needed more RAM and speed and battery.
But the gestalt, the whole, was better than any Apple minimalist thing.
Second favourite:
Nokia 7710.
https://en.wikipedia.org/wiki/Nokia_7710
My first true smartphone.
What was good about it:
– Landscape touchscreen smartphone, with a design that made that the primary orientation, although perfectly usable one-handed in landscape;
– Usable physical buttons for up/down/left/right/select in one hand, call/cancel/hangup in the other hand. So, usable in gloves or without access to stylus and fingertips, with tactile feedback, with both hands. Meaning also ideal for gamers;
– Removable battery;
– Last device with the Psion EIKON UI.
Third favourite:
Sony Ericsson P910
https://en.wikipedia.org/wiki/Sony_Ericsson_P910
Very much oriented to touchscreen operation like an iPhone or something.
What was good about it:
– Familiar and accessible to non-techies;
– With a handy physical T9 keypad, covering part of the screen;
– Flipped down to reveal a bigger screen and tiny QWERTY keyboard. Not much use but clever.
The problem was always too much money (especially on the Nokia side) and not enough user focus. In the Nokia case that meant Series 60, which was where Nokia indulged their most extreme design oddities, while the non Symbian based Series 40 was actually severely good. (North American S40 devices were excessively modified by local operators though, so you would have no idea these things were actually usable if that is all you ever experienced).
If anyone made a decent Symbian device (and supposedly someone in Japan did, of course) I never saw it.
The funny thing is Android launched in an even worse state, but they managed to iterate on it far more effectively.