Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I'm working on a machine learning project where we're trying to package something up, and bringing in all the packages that get installed by default with the framework we're using adds gigabytes to the size of the application, almost all of which is probably not necessary. The challenge is it's bloat on top of bloat, without an obvious way to pare it down, so it becomes a separate engineering effort to try and make something small and efficient. Under a certain size and for many applications, it's more cost effective just not to worry about it and keep the bloat


I found myself in the same situation a few years ago. We hacked together a dubious version of treeshaking using 'python -v' and got the thing small enough to run on an aws lambda which saved a bunch of money. Fun times.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: