You’re working with dynamic languages; it’s a hard problem.

The jetbrains tools do a pretty great job with typescript and when you use annotations in pycharm.... or anything where the types can be rigorously determined; but, obviously, it’s not perfect when the type system says “anything is fine”.

As for building the project; Python just totally sucks at reproducible builds. I don’t think pycharm can fix the entire packaging ecosystem.

So, fair... it’s not perfect; but compared to what? Have you used tools with better support, better stability? Better tooling?

It’s a pretty high bar to clear to say the editor isn’t completely perfect.

Yeah yeah, it uses Java, it eats your ram and the cpu can go crazy, sometimes you need to reset the cache or delete your entire .idea folder.

Eh.

I also wish it was better at some things, but I still love it; what’s the alternative? Visual studio? Hahaha~~~~

And I remember how impressed I was many years ago when I first tried their Rubymine. So magical it made Ruby feel almost like a typed language! :)

> You’re working with dynamic languages; it’s a hard problem.

It's not that hard a problem. At least TypeScript is statically typed and in Python we're using type annotations everywhere. Besides, the bugs I was talking about mostly concern imports (I move file.py from A/ to B/ and PyCharm forgets to update `import A.file` to `import B.file`), and renamings (PyCharm suddenly renames some random variable deep inside dependencies like Tensorflow, too, so I end up having to re-setup my entire Python environment afterwards).

The thing is: Those auto-suggestions for imports work totally fine and code navigation via Ctrl+click works fine, too. So PyCharm does interpret the import statements correctly and I'm at a loss as to why the refactoring doesn't work then, too. Maybe auto-completion and refactoring are two completely separate systems?

I’m not gonna deny it’s frustrating to see quality of life stuff like “don’t refactor stuff from packages!” unaddressed year after year.

...but I think it is a tricky problem.

For example Black[1] is a very carefully crafted project that does an excellent job, but it fails at refactoring files just to apply lints with “inconsistent code generated” on some projects and files; and that’s just apply limiting rules.

...but, I for whatever reason, the JetBrains refactoring for static languages (c#, Java, kotlin) does seem significantly superior.

[1] https://github.com/psf/black

> can't configure my IDE deterministically through text files

Yes, you can[1]. It will even sync automatically with whatever git repo you store the settings in.

> regularly results in situations where building the project works fine for colleague X but not for me

This is a problem with relying on the IDE host to also be the host of your applications. Your repo should contain build settings, not your IDE.

You should switch to Docker or Pulumi or [insert declarative build tool here] -- something more deterministic and consistent across users.

> I'm dealing with bugs in their refactoring engine

As others have said, this is much more likely to be a problem with dynamic languages. I haven't done much with Python, but I've never had a TypeScript refactoring problem in 3+ years of using it as my primary language and IntelliJ as my primary IDE.

> And don't even get me started about CPU usage…

I have a much bigger issue with memory usage, but it's still not a big problem (and I think it's a memory leak in a single plugin anyway).

Also, JetBrains seems to have (correctly) assumed that dev time is much more expensive than dev hardware.

If I have someone whose slow IDE drains 1 hr of work per week and costs $75/hr, then it's only going to be a few weeks before it was more cost effective for me to upgrade her CPU than to worry about the IDE at all.

1. https://www.jetbrains.com/help/rider/Sharing_Your_IDE_Settin...

> Yes, you can[1]. It will even sync automatically with whatever git repo you store the settings in.

No you cannot. Note I was being very precise in my wording. While you can now export your Jetbrains IDE's config as a git repo (or zip file) of text files, you still cannot configure the IDE through text files (like in SublimeText) because 1) the text files' format is not documented anywhere and likely to change in the future without further warning and 2) Jetbrains also writes changes to the git repo automatically. I want to be in control of my config, though. (Sure, I could roll back any changes but I think that's very sub-optimal solution.)

> This is a problem with relying on the IDE host to also be the host of your applications. Your repo should contain build settings, not your IDE.

Well, yes, we do have all our build scripts etc. in our repo but that doesn't keep people (myself included) from running and debugging things from within their IDE because it's a lot more comfortable.

> Note I was being very precise in my wording

I noted your wording, which was not as narrow as your intention. You just said you wanted deterministic, text-based config, which JetBrains has.

I've used VS Code's text-file config, before there was a GUI for most of the config options, and I actually found it to be pretty frustrating. A good GUI is a lot better than editing text, even with the IDE helping you stay within the schema.

JetBrains IDE's probably fail the worst in that regard, I agree; but once you have a config you like, it's easy to keep it consistent.

> but that doesn't keep people (myself included) from running and debugging things from within their IDE because it's a lot more comfortable

Again, these aren't mutually exclusive. You can run/debug code in a Docker container[1][2]. The IDE experience is exactly the same (click the Debug icon or use the shortcut, set breakpoints, whatever), it's just that the VM is running your code instead of your dev environment.

1. Typescript - https://blog.entrostat.com/debugging-a-typescript-project-in...

2. Python - https://www.jetbrains.com/help/pycharm/using-docker-as-a-rem...

I haven't used PyCharm or WebStorm (you mention Python & Typescript, so I guess you're using them), only Rider and peripheral products like dotMemory and dotTrace. I mainly use Rider on Windows, but occasionally use it on MacOS too.

I've encountered the odd issue with EAP versions of Rider (to be expected of course), but the release versions have been rock solid. Coming from Visual Studio, it's been a godsend!

You may just need to tweak your cache & RAM settings. I had similar issues here, and I just juiced the RAM to 4096 (that's 4 GB, a number I chose arbitrarily). Haven't had an issue since.