That's exactly my problem. I really, really wanted to get a Raspberry Pi but I didn't know why (no idea what I'd do with one) so I never did.

EDIT: pimathbrainiac, there's a $25 model.

Quote

Lionel : Yes, my plan is to become a pro. So far, I tried C, C++, Java, Lua, TI-8x Basic, Axe, z80 ASM and HP40g Basic.

Good

Quote

I don't really like the Basic languages nor Lua because they're too limited.

The BASIC languages are, indeed, limited. Their goal is ease of use, not computing power / computing efficiency. What's more, calculator BASIC implementations (and the underlying native code routines) are lackluster, be it on the TI-Z80, TI-68k, Nspire or Prizm - they're slow beyond the cost of being high-level interpreted languages, wanting to deal with multiple data types / performing simplifications, etc.
Lua isn't that limited, though. Its popularity is rising, thanks to it being an efficient interpreted language, being easy to program, and being JITable with LuaJIT (though it remains much worse than native code for heavy computation, but nobody in their right mind does such computations in languages other than, basically, C, C++ or Fortran).

Quote

My opinion for Java is more balanced because it's very similar to C++, but VM and no low level capabilities.

Yeah, but it's one of the most widely used languages. Pretty much every professional software developer has to know a bit of Java.
In this day and age, it's also important to do a bit of JavaScript, and know about the frameworks that are all the rage nowadays (Node.js, Backbone.js, etc.). Again, execution efficiency is not the utmost goal with JS; ease of coding is. With those frameworks, it's possible to build nontrivial HTML5 Web apps in a very low number of SLOC.
Another important thing, in my mind, is to spend time reading about trends and technical solutions on the Internet, and using critical thinking about them. The software engineering field is mutating, thanks to the availability of good open-source components: though it remains important to have programming knowledge, it's also important to know the strengths and weaknesses of a wide variety of solutions, and to be able to assemble ready-made pieces instead of reinventing the wheel. Large companies milking other companies, and state agencies, over custom software that is duplicated or exceeded by open source software, are going to have to change their business model ^^