Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Performance and telemetry analysis of Trae IDE, ByteDance's VSCode fork

github.com

Post reply on HN
Print view View original

261–270 of 390 posts

Re: Performance and telemetry analysis of Trae IDE, ByteDance's VSCode fork

#261

Post by ameliaquining »

cuuupid wrote: post #156
spyridonas wrote: post #22

Great analysis, well done ! Since you've already done VSCode, Trae, Cursor, can you analyse Kiro (AWS fork). I'm curious about their data collection practices.

Anecdata but Kiro is much, much, much, much easier to put through corporate procurement compared to its peers. I'm talking days vs months. This is not because it is better and I've seen no inclination that it would somehow be more private or secure, but most enterprises already share their proprietary data with AWS and have an agreement with AWS that their TAMs will gladly usher Kiro usage under. Interesting to disti…

This seems perfectly rational. If you're already entrusting all your technical infrastructure to AWS, then adding another AWS service doesn't add any additional supply-chain risk, whereas adding something from another vendor does do that.

Re: Performance and telemetry analysis of Trae IDE, ByteDance's VSCode fork

#263

Post by godelski »

throwaway83977 wrote: post #246
godelski wrote: post #148

Earlier quoted context omitted.

> I install vscode from scratch, install a few extensions I need, set 3 or 4 settings I use regularly, and bang in 5 minutes I have a customized, working environment catered for almost any language. Weird, I'd say that's my experience with vim. I just carry around by dotfiles, which is not that extensive. Hell, I will even feel comfortable in a vi terminal, though that's extremely rare to actually find. Usually vi is…

3MB of configuration to make a text editor usable is a lot .

3MB? Good god, that'd be ridiculous!

You misread. I'm using 74K for *vim* configs. (Mostly because I have a few files for organization's sake)

I rounded up to 3M from 2.3M and 1.4M of that is .git lol. 156K is all my rc files, another 124K for anything that goes into ~/.confg, 212K for my notes, 128K for install scripts, 108K for templates, and 108K for scripts

I'll repeat myself, with the *same emphasis* as above. Hopefully it's clearer this time.

  >> The git folder with **all** my dotfiles (which includes all my notes) is just 3M
I was just saying it's pretty simple to carry *everything* around, implying that this is nothing in comparison to something like a plugin or even VScode itself. I mean I went to the VScode plugin page and a lot of these plugins are huge. *All* of the plugins I have *combined* (including unused) are 78M. The top two most installed VSC plugins are over 50M. Hell, the ssh plugin is 28M! I don't have a single plugin that big!

Re: Performance and telemetry analysis of Trae IDE, ByteDance's VSCode fork

#264

Post by kindkang2024 »

raverbashing wrote: post #154

Why would anyone use a "Bytedance VSCode fork" is beyond me

I’m one of those who use it—mainly because it’s cheap, as others have mentioned. I wish Cursor offered a more generous limit so I wouldn’t need another paid subscription. But it won’t. So Trae comes in — fulfilling that need and sealing the deal. This is what we call competition: it brings more freedom and helps everyone get what they want. Kudos to the competition!

I'm not defending Trae’s telemetry — just pointing out the hard truth about why pricing works and why many people care less about privacy concerns (all because there are no better alternatives for them, considering the price.)

By the way, for those who care more about pricing($7.5/M) — here you go: https://www.trae.ai/. It’s still not as good as Cursor overall (just my personal opinion), but it’s quite capable now and is evolving fast — they even changed their logo in a very short time. Maybe someday it could be as competitive as Cursor — or even more so.

Re: Performance and telemetry analysis of Trae IDE, ByteDance's VSCode fork

#265

Post by dmitrygr »

duped wrote: post #252
dmitrygr wrote: post #52

I remind you, again, that vi, gcc, as, ld, and make have no telemetry, launch few (if any) processes, do not need GB of RAM, and work well.

gcc/as/ld are batch processors from the GNU toolchain that offer few (if any) features beyond basic C/C++ (and a handful of other languages) as support, and they're non-standard toolchains on 2 out of 3 major operating systems requiring a bit of heavy lifting to use. It's kind of nonsense to bring them up in this conversation.

And electron is standard tooling? LOL

Re: Performance and telemetry analysis of Trae IDE, ByteDance's VSCode fork

#267

Post by FuturisticGoo »

macintux wrote: post #226
cchance wrote: post #218

Am i the only one that finds System Information: Hardware specs, OS details, architecture Usage Patterns: Active time, session duration, feature usage Performance Metrics: Response times, resource consumption Unique Identifiers: Machine ID, user ID, device fingerprints Workspace Details: Project information, file paths (obfuscated) Not to really bad that obtrusive? Like i don't really see anything there that i'd be o…

I always want the choice to be mine. I was interested in learning Dart until the installer told me Google would be collecting telemetry. For a programming language. I’ve never looked at it again.

It can be disabled btw. And no telemetry is collected on first run.

I keep it disabled for both Dart and Flutter.

Re: Performance and telemetry analysis of Trae IDE, ByteDance's VSCode fork

#268

Post by pmxi »

Aurornis wrote: post #255
kiitos wrote: post #254

In the OP screen share, they toggle various telemetry options on and off, but every time a setting changes, there is a pop-up that says "a setting has changed that requires a restart [of the editor] to take effect" -- and the user just hits "cancel" and doesn't restart the editor. Then, unsurprisingly, the observed behavior doesn't change. Maybe I'm dumb and/or restarting the editor doesn't actually make a difference…

Thanks for watching and catching that. It seems like a major oversight for the core claim: That disabling telemetry doesn’t work. If a restart is required and the tests ignored the restart warning that would invalidate the tests. Either way, it’s useful to see the telemetry payloads.

See the authors response. He or she says it doesn’t matter either way

https://news.ycombinator.com/item?id=44706580

Re: Performance and telemetry analysis of Trae IDE, ByteDance's VSCode fork

#269

Post by tonyhart7 »

user3939382 wrote: post #228
cchance wrote: post #218

Am i the only one that finds System Information: Hardware specs, OS details, architecture Usage Patterns: Active time, session duration, feature usage Performance Metrics: Response times, resource consumption Unique Identifiers: Machine ID, user ID, device fingerprints Workspace Details: Project information, file paths (obfuscated) Not to really bad that obtrusive? Like i don't really see anything there that i'd be o…

I don't want any program on my computer including the OS to make any network calls whatsoever unless they're directly associated with executing GUI/CLI interactions I have currently undertaken as the user. Any exception should be opt-in. IMHO the entire Overton window of these remote communications is in the wrong place.

Yeah but this is industry standard

not saying this is good but everyone do this

Re: Performance and telemetry analysis of Trae IDE, ByteDance's VSCode fork

#270

Post by guessmyname »

raverbashing wrote: post #154

Why would anyone use a "Bytedance VSCode fork" is beyond me

> Why would anyone use a "Bytedance VSCode fork" is beyond me

Because the person using it works at Bytedance.

I guess your question is better phrased as: “Why would any non-Bytedance employee use Bytedance VSCode fork?”, to which I have no answer.

Post reply on HN