News for hackers. Stuff that matters.
Re:James Webb Telescope will run a proprietary JS interprete... by imtringued on Wednesday April 24, 2019 @11:45AM (#19739552)
I completely disagree. It is physically impossible to be more productive in a C++ project that takes 5 minutes to compile compared to almost any interpreted language with a debug cycle spanning seconds. It's that simple. The next issue is the ecosystem. Often the language itself is completely irrelevant if the ecosystem is good enough but here is the thing: ecosystems are tied to languages. Yes I can be productive in "crappy language X" if I recreate the superior ecosystem of "super language S". I am not a superhuman lifeform that has the power of hundreds of bright human brains (millions of bad ones aren't enough) and the corresponding two hundred hands required to interface with the computer.

No. I will just pick a good ecosystem and the language it was built around and be happy instead of proving that you can build a skyscraper out of feces by producing steel via nuclear fusion.

Re:James Webb Telescope will run a proprietary JS interprete... by Jtsummers on Wednesday April 24, 2019 @12:04PM (#19739782)
It's very possible, but instead of debugging you spend more time thinking so you're making fewer iterations.

That said, I really enjoy Common Lisp for being compiled (good performance versus interpreted) and the speed of compiling (compiling single functions at a time) so you get the same time-to-execution as your interpreted languages. Better, actually, if you change the function definition in place when you hit a condition.

EDIT: I hit submit before finishing my thoughts.

C++ isn't that bad. 5+ minute compile times can be tightened down if you have a good structure for your codebase. And if it remains at 5+ minutes, that's what unit testing is for (shouldn't need the whole build to execute) along with a build pipeline. Submit the change and get results after a few minutes. Distcc and other systems can also aid for the really large projects. Another factor to consider, to test that interpreted (presumably dynamically typed) program you have to have more tests than the C++ program needs. Static typing (even though I have issues with C++'s type system, it's still ok) covers a lot of the areas that you have to manually create tests for in dynamically typed languages