Hacker News Reader

Comment by rowls66 | Hacker News Reader

> Scripting languages allow business logic to be written faster and cheaper. The scripting languages (Python, Ruby, Lua, Shell, Perl, Smalltalk, JavaScript) are pretty similar.

That is a pretty bold statement to that scripting languages allow business logic to be written faster and cheaper. I think that it depends completely on the size and complexity of your business logic, and on whether you factor in the cost of maintaining that software over time.

View on HN

Agreed. I've spent a whole lot of time hopelessly optimizing Python so that it ran only 10% the speed of a naive Go implementation (and naturally the Python version was less maintainable as it was a mess of numpy and multiprocessing). Similarly, when I write Python, I spend a lot of time trying to get accurate type information (I've never worked in a project that actually used mypy, so type annotations could be missing or incorrect).

I'm sure dynamic languages were more productive than 90s-era Java, C, and C++, but I don't think those productivity claims hold today.

Please sir don't blame Python for a PhD's NumPY mess.

Well a naive Python solution would be much slower than the Numpy one, so...

Sometimes this is true, and the parent's criticism is certainly unhelpful. That said, I've definitely seen people try to optimize with numpy and end up with something that's even slower (Numpy isn't a substitute for mechanical sympathy, but it's often treated as magic dust by the Python community). Basically Python performance is just a hot mess all around. :(

I guess I'm the OP. Sorry my comment didn't help you; please let me know what you need.

>Basically Python performance is just a hot mess all around.

Maybe don't use Python where performance matters? Just a thought.

That's the right answer, but it's not helpful for people who don't get to make the call on what language to use. Moreover, a lot of people don't think an application will have a performance problem until suddenly it does (either because of scale or shifting requirements or whatever). Even worse, a whole lot of people simplistically believe that you can just throw C/multiprocessing/numpy at any Python performance problem, which is how you make Python even slower and less maintainable.

I'm no great fan of Numpy, but other Python performance solutions aren't more maintainable. :(

> whether you factor in the cost of maintaining that software over time

In many cases the largest cost indeed

Reply on news.ycombinator.com