Try watching someone who is an expert at Cocoa/XCode and you will see that they also have extremely tight, natural interaction with their toolset. They know where the rough corners are and avoid them. Their environment is already configured so that builds just work. etc.
Those of us who work in the browser have similar tools and workflows so that, once we have them working well and meld with the tools, we work very productively. Yet any time I have to show someone new how I do my work, it involves many layers of learning over the course of months before they can get to that point.
Similarly, though your workflow seems quite natural to you, I doubt it would make any sense to me. Yes, I know git and vim, and I can probably read the python code just fine. But knowing how to build the project, how to test incremental changes, how to debug and do performance profiling, and how to have a basic feeling for when I'm done? It would take me quite a while pairing with you before I felt comfortable.
What's the lesson? While there certainly are better and worse development environments, if you judge the one you know well against the one you know poorly, you will always think that what you know is best. I appreciate the sentiment that says learn one thing and learn it well, rather than chasing every fad. But I also think that, if you're going to be well-rounded or wise, you need to have the perspective and take the time to go deep in other environments before judging them.
Even with Python and the Unix tool chain, you still have to learn new libraries and look up API doc when they come out, unless you plan to keep doing the same old things.
Software career is a process of learning new stuffs from time to time, expanding the comfort zone.
I completely agree that there is no common denominator when it comes to build scripts. There are so many variations out there. But I think if I know make, automake, CMake and Ant I'm fine with many projects and some more when I ask the authors nicely to provide me either of these.
But it's not really a problem, actually. Often these people who work in these environments don't understand what freedom of mind is anyway (like try to tell an Apple guy what a FOSS OS would do good for them). The need to implement the whole Unix philosophy and all it's tools didn't stop the Java world from avoiding Linux and BSD which already has the Unix philosophy and tools available. They wanted to develop everything themselves again (and sometimes did a brilliant job!) because that's how they are. I don't like that. I am lazy and don't want to develop what is actually already there and I am not so smart. All my tools would suck compared what already exists. So I am actually quite happy not to work in the Java world even though that's what I studied in university. When git came out I learned how to use it (so much that I would be able to write my own git now), although in school we always used SVN. Now I can happily merge, rebase and cherry-pick and can focus on coding because everybody uses git. Learning git even made my coding better because I learned some great things about software development there.
(This now became more of a novel than a response to your text. Sorry!)
Ironic how the choice of Java and sandboxing was meant to make this not necessary
If you are running the same application on equivalent hardware at the same API level then you should be fine.
Android has issues with hardware as well as the various API versions. The closest parallel in Java would be comparing a Pentium 4 running Java 1.3 vs some 12 core server running Java 1.7 or something.
So a closer parallel would be that 12 core server running Java 1.7 can have its UI skinned by different device manufacturers, and that the UI actually makes a great deal for acceptance of the app in the marketplace.
This is great, but if everybody was doing like you we wouldn't have mobile apps, or in fact anything that has to run on something non standard.
"The Zen of Unix and the Tao of Testing are my companions. Nothing disturbs my chi."
Good, then keep doing your chee. The rest of us are going to use the right tool for the job, instead of some mythical pain-modifier that you think is the "One True Way".
For iOS, however, it seems you're indeed kinda screwed - you have to go through Xcode and the atrocious storyboards stuff.
The main benefit is that you end-up with no merge conflicts in semi-opaque files.
Though I think you're still stuck with xcode ? can you use a CLI toolchain (directly calling clang and so on) to compile, build and sign an IPA ? from the little I've seen of Xcode, it sucks (a lot): random crashes, GUI & behaviour inconsistencies, bad file management, etc...
FWIW, Apple has never required storyboard. You can do everything in code if you -really- want to do it that way, no problem. Personally, I like offloading a lot of the visual stuff to storyboard/xibs unless you start getting crazy with custom animations and so forth.
There is a third party IDE: AppCode by jetbrains which does most of what you need (and apparently lots of things better than Xcode) but i've never used it for any long periods
Xcode gets a bad rap, but i use it every day and can't remember the last time the current version crashed or did something strange, maybe i'm just used to it by now though
What I am trying to say so eloquently (or not) is that maybe the problem is not the modern development world but the developer. The idea that you need just another tool to solve a problem, that is what is flawed. Skill and balance and a set of powerful, basic tools, that is what solves the problem.