Hacker News Reader

Comment by cyberdrunk | Hacker News Reader

IDE has syntactic understanding of your code, which allows you to navigate your whole codebase as if it's one big hypertext document (invalueable especially when learning new code). Also, the syntactic understanding allows you to do one-click fairly complex refactorings with guarantees that nothing will get broken (at least it works for sane languages, like Java. I don't know how well it works for C/C++, where preprocessor macros complicate things). I don't see how "UNIX as an IDE" delivers this.

View on HN

Yeah, UNIX doesn't do that on its own. You can grep, or run find and do a bash loop and so on, which is cool. But it's language servers that give semantic understanding of your code. You can easily refactor, or go to definition from vim. So UNIX is a development environment, very flexible, yes, just not that integrated.

I have mainly used vim for development and I can navigate my codebase just as seamlessly with ctags-generated indexes.

I'd agree on vim being of limited utility when it comes to automated refactoring (it's a text editor after all), but beyond changing the names or extracting functions what can an IDE do for you in this respect?

Refactoring in IDEA also allow you to for example also change function signature, safely delete an element (class, function etc.) - i.e. with full impact analysis, move code, extract interface/trait out of a class.

I've tried the ctags-indexes with Sublime and afair the basic navigation was not bad (I think I remember there being an issue with some delay though? It was a while ago). I'm not sure if things like "show code which access this variable" or "show all children of this class" are supported though?

Apart from those advantanges, I can think of two more:

1. IDE can import your build file (maven, sbt or bazel project for example), parse out the project structure from it and download the necessary dependencies. I don't think that ctags-like powered solution is as robust, which means you'd have to fiddle with it until it can understand the structure a complex repo.

2. IDE can run and show the results of unit tests. (just right-click on test you want and select "Run" from context menu).

My build tool spat out a 700MB compile_commands.json file, which the C++ language server turned into 25k files. None of the language server implementations were able to process even a go to file on the code.

it takes about 15 minutes for Visual Studio (well, visual assist) to generate the index for the project, and then the navigation is instant. I've tried a few times over the last few years with various language server and clients, and my experience has been the same pretty much every time.

You can use ctags/cscope to handle that, FWIW languages like C and C++ have a massive toolset built around Vim, so even for the most obscure use case, I'm pretty sure there will be a way.

> at least it works for sane languages, like Java.

You lost me here.

> I don't see how "UNIX as an IDE" delivers this.

Have you read the linked series of articles? The author explains and provide examples with C and Perl, Iirc there's a book about working with Makefiles in Java, so it's totally doable, painful, but doable.

Reply on news.ycombinator.com