IntelliJ IDEA 2023.2
jetbrains.com
jetbrains.com
The time it takes to start a Java program in a Maven project is excruciating. And for unit tests or small utilities that's something I need to do countless times a day. While some random code analysis feature really is maybe helpful once a year. I actually reverted back from the last release because it got so bad.
Am I the only one with this problem? Has anybody found a solution for it?
And the new re-write they started looks like it isn't because they really wanted to do it, but because they felt a pressure to do something about VS Code: https://www.jetbrains.com/fleet/
As for starting a Java program in a Maven project - are you sure that's IDE overhead or is it also slow from the CLI?
JetBrains IDEs have this problem all the time and its not consistent across all platforms. i.e. some people see it, some don't.
At this point I wish I could switch off JetBrains but I'm so familiar with it and using VS Code my productivity just screeches to a halt.
I've gone through this process from vim > eclipse > komodo > atom > vscode
now that's a name i haven't heard in a long time...
I was also initially very skeptical of the new UI, thinking it was purely a VSCode imitation that would remove power tools, but after using it for a few days I was converted - all of the power I used was still there, but now with a simpler and more efficient view (less clicks to find what I want, etc.).
Only started using IntelliJ about 3-4 years back (due to Android development) so I don't have any comparison from older days there.
No change / add in plugin's everything same.
Switched to using '17.0.8 Eclipse Temurin Java Runtime' and it's fast as before this release.
Hope this helps someone.
Maybe they could raising a warning to the user when it sees evidence of lots of GC trashing?
At least for groovy dev (IntelliJ is basically the only IDE you can use for groovy functionally AFAIK), it starts to really slow down once I get to 50-100 classes, and starts to trend to unusability. There's definitely a nonlinear growth in the impact of more code.
So I need to split my projects into sub-jars, a bit of a PITA.
Now that I work primarily on Java and Kotlin projects, I’m using IntelliJ, where the entire IDE is subject to the JetBrains effect. The only tactics that keep me productive is to significantly delay updates, keep plugins to a bare minimum (the fact it is not very rare for plug-in updates to brick your environment is really frustrating), a losing battle of identifying and disabling unnecessary functionality, throwing more hardware at it, or restructure large projects into smaller bits and import as modules. Rapidly approaching the point of it not being worth it, but I hate pretty much all the Java IDEs I’ve tried.
But if you do give it like 2GB of RAM at the least, it is very smooth (unfortunately it used to have a very low max memory setting for a long time). Java is good at handling dynamic loading, you don’t really pay for any feature you don’t use, they don’t even get class loaded.
1. Make sure it's not your machine itself that's getting slow. On Intel Macbooks try running "watch pmgset -g therm" and watch the speed limit. If it drops below 100 then nothing you do to your IDE will help.
2. If you're seeing freezing, try enabling the memory indicator in the status bar. If you see the freezes correlate with drops in heap usage, increase the maximum heap size. In this release they increased the default to 2G but that's still really quite small. I have it set to 8G and never see stalls.
3. On macOS try turning off "Displays have separate spaces" in System Settings. You can also experiment with toggling ProMotion off and setting the display refresh rate to 60 Hz.
https://youtrack.jetbrains.com/issue/JBR-526/IDE-is-unusably...
4. Try turning off "matched brace", "show whitespace" and "show code lens on scrollbar" in settings.
A lot of performance issues these days are to do with macOS and the myriad different ways you can fall off a performance cliff especially when working with external displays.
This command (with -g therm) shows CPU thermals. You can also run top or Activity Monitor and if you see kernel_task taking up a lot of CPU, that's the OS throttling the CPU for heat.
I'm also very interested to hear what their billing will be for this, which they conveniently don't mention.
...because, bluntly, that openAI sh$t isn't free (specifically custom models), and either a) jetbrains is taking a hit to their bottom line using it, or b) they're going to roll out new pricing off the back of it.
This language clearly indicates possibilities for local models in the future or at the very least other options for hosted models.
Would you rather them roll out cost-plus pricing or roll up the cost of new features into higher base prices?
I don't understand what the exact problem is you're having.
I could understand your point of view if it was a feature that could be made free in some way. But it literally can't - so you just don't want others do have this feature, because you can't have it for free. Why?
Code gpt allows you to select code and then trigger different prompts for that. Like review this code, find bugs, optimize this code, etc. If you select gpt 4, it gets pretty OK. I've caught a few edge cases with it.
I appreciate the community version of IDEA being open source. I was able to look up the implementation for how java that handles this. The c++ implementation I cannot access. I've added some comments with my results from a profiled run on some old tickets about this (including one as far back as 8 years ago) but it doesn't seem to be a priority, yet!
I had the genius idea that you should be able to read the error messages, especially when it was related to types. The previous output was almost unreadable.
They finally agreed to add it after someone built their own plugin to do the same thing in VSCode.
A bit frustrating that it took this long, but glad they did it. Over the years, I've tried every IDE on the planet (I have no religious notion of the best one), and this one remains my favorite simply because it works very well across a lot of languages and because they do listen to feedback in their issue tracker (eventually).
Their quality has gone down hill lately. Plugins are a bit half baked or missing lots of functionality. They for the longest time would eschew using the community available LSP as opposed to trying to build their own, they finally capitulated on this (I saw this with the Vue related updates in Webstorm that they're using Volar now, for instance)
I know they're likely throwing alot of effort behind Fleet, but its taking way too long to get out the door as usable. Its missing too much. Its a chicken / egg problem in that I can see how Fleet is the next generation of Jetbrains but its not far enough along to supplant any of their IDEs because its missing too many features and plugins.
I really hope this is a lull (all companies have them) and they come out swinging again soon
Lack of proper LSP support is killing them and their recent development isn’t enough
As for Fleet... it's shit. It's just a slightly nicer VSCode but I have zero interest in editing JSON files to configure my plugins. I think in the end it's just going to be a less popular VSCode. I see it as a bad distraction for JetBrains and it worries me quite a bit. Stop chasing VSCode and improve your IDE.
Also I am using PyCharm 2022.1.4 (Professional Edition) because they removed the functionality to edit/customize ssh interpreters in the later versions.
I love PyCharm, as I work mainly with ssh interpreters my patience is tested everyday
Really love jetbrains IDEs.
-> View -> Appearance -> Navigation Bar
I am sure some genius designer can make it work, but when even Apple fucks it up, I don’t want someone who tries.
I can’t remember the last time I got flamechart navigation to work in WebStorm. How about fixing what is broken before adding new, guys.