I mean I could buy it if Python was maybe 5-10x slower than "fast" languages, but the benchmarks people are throwing around show it is 50-100x slower. Every Python codebase I have used (apart from one-off scripts I guess) has eventually run into the "ok it's slow, how can we make it faster?" barrier.
Now does it matter? Not all codebases are the same. If I'm serving api requests, does it being 55ms vs 15ms make a difference? 110ms vs 75ms? If you're writing analysis on a huge data set and an iteration takes 100ms vs 50ms? yeah that could be the difference between weeks vs days.
Almost like someone should look at what they're doing before saying 'IshKebab said python is slow, so we should rewrite all our code'
But usually 90% of the response time for the api request is the time it took the SQL server to execute the relevant queries.
I make software that requires searching over large datasets (image recognition for construction drawings). It's a web app running python on the server and it feels instantaneous for users as they're searching.
Most people aren't even doing that - they're just pushing and pulling data from a db. Building maintainable software is what really matters.
Yes, it's the difference between being able to serve 66 r/s or 18 r/s.
That's why I'd never start a new project in Python (among other reasons). It backs you into a poor performance corner.