I learned Python and it made me enjoy writing software again. 8 years later and it's nearly the only language I use. In principle it makes more sense to choose the right tool for the job, but if I don't enjoy doing the job with other tools, it makes sense to plan around the tool.
I avoid mobile or front-end web development because Python isn't the right tool for those jobs. Fortunately, I like server-side development and data engineering and Python fits perfectly there.
Whatever weaknesses Python has either don't affect me because they don't intersect with the way I work, or else I easily work around them in ways that don't feel like compromises to me. But I can understand how someone else could say the same things about a different language.
With Python you need to choose a framework, choose a wsgi server, and hook all that together with nginx.
First of all, is this a solo project or is a group of people going to be working on it? If it's a solo project, then you are far less constrained. If it was a small proof of concept or otherwise throwaway project, I'd probably use it as an opportunity to try a new language. If it's a real deliverable and doesn't involve hard real time performance, I'd choose Haskell because it's a language with which I'm proficient and I enjoy using.
If this project is going to be developed and maintained by more than just myself, then the story changes. With an existing group of developers, you have to play to their strengths and get buy-in, so use either a language they already know, something similar, or something they've shown an interest to learn. If you don't yet have developers, consider the difficulty of hiring for certain technologies. There are also performance and correctness requirements to consider.
If you happen to be an enterprise decision maker (which is unlikely since this is a community for intelligent critical-thinkers) then the only correct choice is Java. No one gets fired for choosing Java. There's a framework for everything. There are plenty of cheap developers. It's popular, so it must be good.
Want to build an operating system in Ruby? Ok, good luck trying to pour concrete with your tablesaw.
Admittedly, that analogy is a little extreme since programming languages are Turing complete, etc etc. However, if that's your perspective, you may want to have a gander at Cobol on Cogs: http://www.coboloncogs.org/INDEX.HTM
2) Does it suit the nature of the problem?
3) Does it have a large enough ecosystem to address my/our needs?
4) If it doesn't, do I/does the team have enough time to fill in the holes?
5) Can it be deployed in a reasonable way?
First, what is the type of project? If desktop, then C#/WPF (since my desktop apps will always be Windows). If web, then the choice becomes more complex.
For web, am I going to be solo? If so and there isn't a compelling reason to pick a specific language (be it because of its strengths in a specific area or environment constraints) pick whatever language I feel like learning or already know (if it needs to be rushed then pick a language I know).
If I'm doing web and going to have a team, then pick something that I find interesting, that fits the needs of the project and I can find people if need be. I'm currently working on a project that is just me but will eventually be a team of people (somewhere down the road). I chose Elixir/Phoenix because it is something I want to learn and there is no specific reason for picking some other language. I've heard enough about Phoenix and Elixir to know that I could find some Ruby/Rails devs to teach them Elixir/Phoenix if I wasn't able to find elixir devs, so making a team isn't a big deal.
Does the project involve certain functionality that's already been solved in a framework or library available only in one language (or at least not in any languages that my team and I are already proficient in)? Will it be faster for me or my team to learn that language thoroughly than to reimplement the functionality?
Is the project being implemented on a platform that only supports a limited set of languages officially (e.g. iOS or client side web development)? Pick the most widely used of those.
Otherwise the project gets done in the language I enjoy most and am most proficient in (for personal projects). Because that'll be the most fun for me.
Or the language my team collectively knows best and were hired for their knowledge of. Because that makes knowledge transfer and future maintenance easier.
Else if it's for work: Java and/or JavaScript
Else Python
- How complex is the thing I wanna do? If it's very complex I'll go with Python since I'm much stronger with Python than Go. This allows me to focus on the problems I'm trying to solve not the ones I create by misunderstanding something in Go.
- What are going to be some of the key features I'll need from my tools? Concurrency is not a strong point of Python so something that requires it may be better off in Go. Am I going to need auth, email sending, an ORM? Django is great for getting that crap out of the way so I'll probably use Python.
- How much new stuff will I be learning for this project? Most of my personal stuff is done as a learning exercise. I have found in the past I tried to learn too many things at once and I would just end up lost.
These are just some of the things I think about before choosing a language. However, your milage may vary of course. What ever questions you ask yourself you should always keep in mind the strengths and weaknesses of your options and whatever you do end up choosing you keep those strengths/weaknesses in mind and work to make the most of the strengths and use other tools/processes to mitigate the weaknesses.
Desktop or Web? Data, Graphics, Game or something else? Who is the audience? Need to be expendable? Platform specific or agnostic? Input/output considerations? Single user, centralized or distributed data? If Internet/network what sort of connections will you need to support AS A MINIMUM? If planning to use commercial tools, can the audience handle the distribution/licensing requirements, costs? Update considerations - if a closed source single user, consider how updates will deploy in the platform you choose. Any special hardware you plan to use (i.e. pinpad, signature digitizer, fingerprint scanner) these will be a BIG factor on what you choose, sometimes such devices will narrowly define what is usable.
Many of these are not much a problem by using more tried and true languages/platforms than something new as many of those issues have solutions you can just Google an answer. Going toward the newer/shinier platforms might get you an edge in some feature but may completely ignore other factors.
The reflection part I could have gotten around fairly simply, but I'm happy with how cleanly it worked with reflection. Polymorphism was a must-have, and garbage collection was close (it might have doubled the amount of work if I didn't have it).
I don't find reflection itself to be particularly helpful, e.g., the only times I used it in Java were during low-level framework creation and code generation. Anything else is better handled via interfaces or similar functionality.
But if I were forced to use the JVM (which is a nice platform) I'd tend towards Clojure (distinctly not OO). If OO was a strict requirement then Groovy or Scala.
Java's OOP implementation, however, is fairly limited in scope, although Java 8 has somewhat alleviated my own particular pain points (e.g., lambdas).
Use Scala for building systems that actually require the OO structure. Use Apache Groovy for glue code and tests, and other scripty stuff like build scripts for Gradle.
I've used Perl and Python for small applications and glue code. I liked Perl for a long time, but I don't want to go back to it after using Python.
Most of my personal projects are written mostly in C++, sometimes with bits in C, because they're suitable for emulators, game engines, renderers, Arduino programming, and robot control code and because (as previously stated) I like working with them. I've been meaning to start some things in Rust, but haven't gotten around to it, in any serious kind of way.
I choose employment based partially on what language they're working in. Systems-level stuff on Linux? Probably C or C++, so I'll probably be happy.
1. Your time
2. Machine time
3. Others time
Balance those three to find what works. Need really fast code? Sacrifice your time and co-workers time. Need really maintainable code? Sacrifice your time and machine time. Need really quickly written code? Sacrifice machine time and co-workers time.
You can keep it simple.
2) use java
Otherwise C# is easy and reliable. Less secure than Rust, but less of a concern if it doesn't face the web.