Programming code is more often executed than it is read or written.

The software world tends to claim that code is read more than it is written. By that we mean that code is mostly written once, but other devs who maintain it, or who need to find bugs need to be able to read it in order to understand it.

This is why Perl has such a bad rep: It can produce code that is terrible to read, due to the Unix shell and awk influenced “Magic” that can be used to make code shorter and faster, but also close to impossible to read at first glance.

But the focus on readability should not stop us from looking beyond development. Take interpreted languages, such as Javascript, Python and PHP. While I completely agree that these languages are simple to read, I am appalled that al this code is interpreted every time the code is executed, just in order to make life sometimes better for developers.

Most interpreted code is run ‘as-is’ at customer premises, and will never ever be examined. How many WordPress users will ever examine the base code of WordPress? How many people browsing the web will ever examine the React Javascript code? You know the answer: None.

It is absurd that we keep this code uncompiled (to at least bytecode) just so that once in a zillion requests a developer might want to check the code. What is good for the developer is is this case clearly not good for the user.

PS: Yes, JIT can help mitigate this to some extend, but is still an order of magnitude worse than a compiled language. Also: almost all JIT interpreters do not have a persistent compiled code cache spanning multiple sessions. They are made to reduce latency within a session, not to reduce resources in the lifetime of the software usage.