[coderunner]

What about in the context of trying to get to an MVP? Is the dev time speedup of using a dynamic programming language and stack significant over using a c++ backed? You wouldn't care much about performance when you're trying to figure out if you'll get traction.

[otabdeveloper4]

No, it's not significant. Dev time depends on programmer skill, not the toolset. A good C++ programmer will develop your MVP many times faster than an average Python programmer.

Python programmers are much easier to hire, though - you already need a good C++ programmer on the team to hire another one, because HR and corporate management can't into proper hiring process.

This last factor is the overarching most important one for BigCorp Enterprise Inc., not development speed or cost.

[filleokus]

> Dev time depends on programmer skill, not the toolset.

This is obviously not strictly true, always. A skilled programmer will use the proper tools for the job.

If you for example is tasked with writing a backend service exposing a GraphQL API, I think it would be foolish to do this in C++, and would bet that the average Python programmer would do it quicker than even a top-tier C++ programmer (if the latter would be hellbent on doing it in C++).

Especially when working with MVP's (or new projects in general), the ability to leverage already existing tools and frameworks are key to rapid progress. This doesn't necessarily have to be scripting languages, but the Python/Node/Go/etc developer would have a working GraphQL server up and running connected to a database of choice within an afternoon while the skilled C++ developer would have to spend at least a few days implementing a GraphQL server mostly from scratch [0].

[0]: A quick Google show that schema parsers exists for C++, but nothing matching the frameworks/library available for more web-fashionable languages.

[otabdeveloper4]

Two points:

a) Writing a schema parser is not rocket science. In fact, for a good programmer implementing their own GraphQL library would be quicker than integrating some third-party library. So your first point ("average Python programmer would do it quicker than even a top-tier C++ programmer") is absolutely wrong.

b) There's no value in an MVP that does something generic that is already available in off-the-shelf libraries. Your GraphQL example is pretty pointless because it doesn't actually do anything.

> ...the Python/Node/Go/etc developer would have a working GraphQL server up and running connected to a database of choice within an afternoon

Well, no. By the end of the week they'll still be arguing about which package manager to use, whether TDD is a good idea, what makes a microservice 'micro' and how to configure Kubernetes.

[stefano]

The same argument can be made for Python (or for most non-trivial jobs, for that matter). You always need a good technical person to hire another one.

[otabdeveloper4]

Not really; hiring Python and Java devs is very amenable to keyword-driven recruitment. Hiring C++ devs in the same manner is a clusterfuck waiting to happen.