DEV Community

edA‑qa mort‑ora‑y
edA‑qa mort‑ora‑y

Posted on

Relationship between programming language and run/compile-time speed

My answer to the question Does the programming language a program was written in affect how much time it takes to compile or how much resources it uses?

Yes to both, but the relationship isn’t simple.

Compile Time

Some languages can be compiled fast, and others are relatively slow. Without looking closely at the language and how the compiler works there isn’t a simple way to predict if a language will be slow to compile, but there are two major considerations.

Structure - The structure of the language itself impacts compile time. Does it need to include many files (like header files), does it resolve a lot of typing (like C++ parametrics), does it have a complex syntax that requires multiple parsing passes, can processing be done in parallel, or must it be serial, etc?

Dynamic vs Static - How much is done at compile time makes a difference, though often not a large difference. Languages like C++ do most of all name and type resolution at compile time, which may increase it’s time, whereas Java does some at runtime, and dynamic languages like JavaScript do most things at runtime. Of course, there are JIT’s involved at runtime, but as the time is spread out it still feels faster.

Resources Used

It depends a lot on what program you’re writing how much the resources will differ between programming languages. So let’s break this into two considerations.

Overhead - The way the language works with memory, objects, functions, and concurrency can introduce overhead. Overhead is stuff that you’ll encounter unless you try hard to avoid it. The more abstract the language the more overhead you tend to have. There also tends to be more overhead in dynamic vs static languages (see compile time). This overhead applies across all the code you write in the language. It may not apply to libraries you use, as they are written in other languages, or have been optimized. This is why it’s not problematic to use inefficient languages in some domains, as they account for only a small portion of the actual code running.

Pattern - The structure of a language, and it’s standard library, heavily influences the way people program in that language. Syntax aside, C++ and JavaScript programs just don’t work the same way. This is perhaps the biggest impact of performance most people will encounter. It also means you can probably use any language in an efficient way, and an inefficient way, if you try hard enough (though it may be an uphill battle).

Not Paradigm Specific

What’s interesting about this is that none of it directly relates to the paradigms that a language covers. The factors could apply to a language in any domain.

And though specific features can make it slow, it’s usually the history of the feature doing that. Complex features can be quite fast, as I’m discovering with my Leaf compiler.

Latest comments (0)