The stuff I write I don't think is that bloated, but like most things these days the stuff I write pulls in a bunch of dependencies which in turn pulls in their own dependencies. The result, pretty bloated software.
Writing performant, clean, pure software is super appealing as a developer, so why don't I do something about the bloated software I write? I think a big part of it is it's hard to see the direct benefit from the very large amount of effort I'll have to put in.
Sure I can write that one thing from that one library that I use myself instead of pulling in the whole library. I might be faster, I might end up with a smaller binary, it might be more deterministic because I know exactly what it's doing. But it'll take a long time, might have a lot of bugs and forget about maintaining it. Then end of the day, do the people that use my software care that I put in the effort to do this? They probably won't even notice.
I think part of it is knowing how to use libraries. It's actually a good thing to make use of well-tested implementations a lot of time rather than re-inventing the wheel: for instance it would be crazy to implement your own cryptography functions, or your own networking stack in most cases. Libraries are good when they can encapsulate a very well-defined set of functionality behind a well-defined interface. Even better if that interface is arrived at through a standards process.
To me, where libraries get a bit more questionable is when they exist in the realm of pure abstraction, or when they try to own the flow of control or provide the structure around which your program should hang. For instance, with something like Ruby on Rails, it sometimes feels like you are trying to undo what the framework has assumed you need so that you can get the functionality you want. A good library should be something you build on top of, not something you carve your implementation out of.
Writing performant, clean, pure software is super appealing as a developer, so why don't I do something about the bloated software I write? I think a big part of it is it's hard to see the direct benefit from the very large amount of effort I'll have to put in.
Sure I can write that one thing from that one library that I use myself instead of pulling in the whole library. I might be faster, I might end up with a smaller binary, it might be more deterministic because I know exactly what it's doing. But it'll take a long time, might have a lot of bugs and forget about maintaining it. Then end of the day, do the people that use my software care that I put in the effort to do this? They probably won't even notice.