Live data from Hacker News

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

github.com

351–360 of 390 posts

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

#351

Earlier quoted context omitted.

> Bash isn't better. It is, by virtue of running on Linux. > because I write and run such scripts. 'kay. Learn how to do Engineering and the software will come just fine. You don't need telemetry to tell you anything about scripts. You need good error reports for your users to send to you instead. > Microsoft wouldn't suffer, PowerShell users would suffer. So what you're saying is that Microsoft doesn't care about it…

[flagged]

> That's good, because if running on Linux is what it takes to make something good, PowerShell is good, and has been good since before you created your HN account.

Is there even a single distribution of linux with PowerShell as the default shell?

No, of course not, because PowerShell does not work on Linux. It does not work with coreutils, it does not work with bash, it does not work in any way that's useful to Linux.

> This does not tell you the thing I said the telemetry was added for. Try again. Or don't but if you're going to, at least understand the claim and address it.

If your servers don't have users, then why are they running scripts?

Your servers have users. The users are you, the server admins, whoever's consuming the data being served by them. Your users need to tell you that something is wrong, voluntarily. If you are not getting reports from your users, then you are doing it wrong.

> If you are forgetting the context of the chat, you can re-read it.

'kay. Neat. Nothing was forgotten.

> That is not what I am saying. A team which is fighting to get more funding so it can address more issues is demonstrating that it does care about users.

> I refer you back to my previous comment where I link to the Github where you can see they are interacting with their users, and wonder why you are immediately saying the opposite and thinking that will fly.

It's because you're too stubborn to recognize that you're fighting for things that harm users.

A team which is fighting to get more funding, and debases themselves to do it, is not a good team. A corporation which fosters that kind of thinking is a terrible corporation.

Showing github issues and PRs and stuff? Cool, good interactions.

Trying to justify telemetry by tying it to funding? No, not good.

Get your github issues status to be tied to your funding instead.

> Were you listening at all, or were you just waiting for a chance to tell everyone you don't know what people do with computers, haven't looked, don't care, but are certain all they need is Markdown and a terminal with a tiling window editor?

I'm not sure what you're getting at.

I suppose Markdown came about because it's extremely simple and great for inline code / readmes when contrasted with all of the bloat that comes with HTML, or contrasted with the binary blob that is PDF. It's great and there are plenty of people who don't use Markdown. How do I know? Because I talk to people.

As for a terminal with a tiling window editor? There are plenty of people who don't use terminals, and plenty of people who do. Have you sat down and talked about how and why? Have you tried to actually have conversations with multiple other developers to understand their workflows?

Because if not, then you're only guessing at why people use a given thing. Telemetry doesn't solve that at all.

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

#352

Earlier quoted context omitted.

As a somewhat paranoid person I find this level of paranoia beyond me. Like do you own a car? Or a phone? A credit card? Walk around in public where there's cameras on every block? I don't agree with it at all but the world we're living it makes it impossible to not be tracked with way more than (usually anonymized) telemetry data.

> (usually anonymized) telemetry data. Anonymization is usually a lie: https://news.ycombinator.com/item?id=20513521 https://news.ycombinator.com/item?id=21428449 Also please stop with security/privacy nihilism, https://news.ycombinator.com/item?id=27897975

It's not nihilism. I still ad block, use an RFID wallet, and don't install any apps on my phone, I rarely use google for anything. But at some point when something is so ridiculously useful and the data they're getting doesn't really mean much of anything I have to stop caring. I use Windows 11 (gross) because it lets me play video games with my friends I can't otherwise. I use Uber because it lets me get across town. I use Visual Studio because it helps me code. I use Chatgpt because it helps me with so many things. To take away any one of those because I'm a privacy absolutist seems silly to me. It has the exact same vibe of never leaving your room because you're afraid of all the cameras.

I'd like there to be a push back against these companies because I find their practices disgusting but running linux with only open source software and a fairphone is just an extreme I'm unwilling to entertain because it's just not possible in my (or most people's) world.

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

#353

Earlier quoted context omitted.

> Bash isn't better. It is, by virtue of running on Linux. > because I write and run such scripts. 'kay. Learn how to do Engineering and the software will come just fine. You don't need telemetry to tell you anything about scripts. You need good error reports for your users to send to you instead. > Microsoft wouldn't suffer, PowerShell users would suffer. So what you're saying is that Microsoft doesn't care about it…

[flagged]

The problem is not the PowerShell itself, maybe even not the fact it has telemetry, but how it collects the telemetry and how Microsoft sees telemetry as the only and ultimate way to measure things.

I understand that Microsoft believes in "show me numbers or GTFO", but the problem is they ignore Goodhart's law repeatedly, and they don't learn from their mistake. To recap; "When a measure becomes a target, it ceases to be a good measure".

So, to keep the line go up, Microsoft's all tools are enshittifed over time, and this metric collection is marred by internal power struggles, confirmation bias, and representation inequality (when a company uses a feature in all their installations because it runs a script developed and distributed by a single engineer, their representation is amplified, for example).

Do you remember how a single person gutted .NET Core to remove hot reload support because of a power struggle, despite the community and the numbers screamed the opposite? This is the Microsoft we know, and since we rank ourselves by our HN account creation dates, you know it way better than me.

In many software communities, it's shown over and over again a distilled stream of human feedback is way better than playing the numbers game with some automated number collection and crunching, because it's way better filtered than some raw numbers. Another problem is Microsoft is not merging the numbers with actual feedback, and going for a hybrid approach either. They are not a company like this. They are not formed like this, and their DNA doesn't allow that. They may be able to do it top down maybe in a decade, but today they're not there yet. Their community response is just fronting the issue trackers and let some feedback trickle to keep the community happy. Yes, I don't believe in or trust Microsoft a bit to care about their users unless it affects their bottom line.

Like it or not, regardless of Gates, Ballmer or Nadella; Microsoft has never changed at the core. It's a company based on IP, suppressive monopoly, and "win at all costs" mantra.

So, at the end of the day, the numbers are not there to secure user happiness, or giving clout to the Powershell team. They unfortunately have to do it for job security, and to have some arm's length distance from the shredder. You today have PowerShell because, some big customers said they need it, team can show it's being used by some other minions, and maybe a couple of internal teams benefit from its existence.

Don't get distracted. It's about money [0].

[0]: https://calebhearth.com/dont-get-distracted

P.S.: ...and yes, Bash maybe a terse language, but it's better than PowerShell, a lot.

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

#354

Earlier quoted context omitted.

To take that logic to its extreme: I'm sure we could have amazing medical breakthroughs if we just gave up that pesky 'don't experiment on non-consenting humans' hang-up we have.

The parent said "talk to your users instead of telemetry" and I said "there are scenarios where telemetry can get information that you cannot get by talking to users". How did you go from that to "experimenting on non-consenting humans"? To take your logic to its extreme, you have a disease and are prescribed pills, and the pharmaceutical company says "we will track when you take the pills - unless you don't want us…

Medical research and consent doesn't work like this. If you track your patients without their consent, or you share their data without their explicit consent, you'll land in very hot water, which will cook you even before you can scream.

Similarly, a medical trial will take a very detailed consent before you can start.

Your opt-out telemetry is akin to your insurance sending you powered and Bluetooth enabled toothbrushes out of the blue to track you and threaten to cancel your insurance if you don't use that toothbrush and send data to them.

Or as a more extreme example, going through an important procedure not with the known and proven method but with an experimental one, because you didn't opt-out and nobody bothered to tell you this. In reality, you need to sign consent and waiver forms to accept experimental methods.

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

#355
post #215

Earlier quoted context omitted.

> Why isn't there a decently done code editor with VSCode level features but none of the spyware garbage? Because no other company was willing to spend enough money to reach critical mass other than Microsoft. VSCode became the dominant share of practically every language that it supported within 12-18 months of introduction. This then allowed things like the Language Server Protocol which only exists because Microso…

well, to be fair, LSP is good gift.

Except that the LSP is now trapped in amber and cannot be evolved because it requires the agreement of Microsoft to change.

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

#356

Earlier quoted context omitted.

Yes, why do people use products from Microsoft, Apple, Google, Amazon, ...

> Yes, why do people use products from Microsoft, Apple, Google, Amazon, ... I work at Apple, so I’m not concerned about being monitored—it’s all company-owned equipment and data anyway. It was the same when I worked at Microsoft. I used Microsoft products exclusively, regardless of any potential privacy concerns. Employees at Google and Amazon do the same. It’s known as “dogfooding” —using your own products to test…

Hmm. Are you aware that I was responding to this comment?

> Why do people use obvious spyware when free software exists?

So, even though the poster was referring to ByteDance when they said "obvious spyware", I was feigning incomprehension in order to ask the question, how do we differentiate ByteDance from what Microsoft, Apple, Google, Amazon (and the rest) do.

It's a real question - why do technical people, who arguably should know better, and can do something about it - continue to use these data-harvesting and user-selling platforms? The answer is obvious when it's the case of an employee of those companies, I grant you that.

My apologies if you feel your response did address that, and I missed it. If so, please help me see what I missed.

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

#358
It’s honestly kind of funny — you got timed out by Discord’s automod and now you’re running around calling it “suppression of technical discussion.” lol. Don't be dramatic. There's a wave of crypto spam before, so the automod timeout often catches lots of bots and people. I remember even community mods have been timed out before.

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

#359
post #99

Why do people use obvious spyware when free software exists?

Yes, why do people use products from Microsoft, Apple, Google, Amazon, ...

If we are talking about telemetry…. Seriously all these products are sending telemetry data

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

#360

How is memory usage related to telemetry? I don’t see a lot of direct correlation here. If they cut down their mem usage to the same level of cursor did, does that mean they are not sending that much data then?

pretty much every serious dev tool collects some form of usage data. I don’t think this is some evil conspiracy. it’s how teams figure out what’s working, what’s broken, and what needs improving.
Post reply on HN