I was also thinking of hobbysts / tinkerers, that make a non-trivial
percentage of python users, perhaps on the way to become professional
programmers. They run some script and they just want to change a line or
Even experienced developers benefit from having an easily installed development environment. We had some of these in the past but they've kinda fallen unmaintained. If we're going to move forward with a project like this, I think that a Lua+LuaRocks distribution in the same vein as LuaForWindows is a good idea.
Also, with the couple of Lua-based open-source applications I published
I found out that most Windows users have trouble even setting a path and
writing things correctly on a command line. So, the easiest, the better.
I am of the opinion that this is a lost cause. Publishing applications should be done with a packager, and this is true regardless of platform. You're not going to solve this problem for non-technical Windows users no matter how easy it is.
I think Python is successful despite its shortcomings because, apart
from the bandwagon effect:
- It can be installed and used immediately. If you need a module, you
just import it.
Has Python gotten better at this or something? Last time I tried it I had a real headache getting Python installed on Windows in a way that wasn't a headache to use for anything but launching IDLE from the start menu.
Python isn't immune to the problem, either; if you go outside of the standard libraries, you still need pip to pull in dependencies and cross your fingers that there's a binary build that'll work on Windows.
On that note: Promoting a beginner-friendly Lua IDE, even if it's not suitable for professional use, is going to be the biggest step we could take towards making it easy for novices to get started with Lua development on Windows.
> It should be pretty straightforward for a LuaRocks Windows distribution
> to include a C compiler that's already configured appropriately. This
> shouldn't be considered an obstacle.
It shouldn't, but I suspect it could not be as easy as it sounds. I'll
be glad to be proven wrong.
The part that's hard isn't including the compiler; it's ensuring ABI compatibility with external libraries. This more or less means that binary distributions are mandatory for anything where you can't just build the entire dependency chain. The solution is going to require module authors with compiled dependencies to go out of their way to ensure Windows compatibility.