Hacker News Reader

Comment by ho_schi | Hacker News Reader

Bugreport for headline

    Linux users provide more and better bug reports. Which helps developers improving overall quality.

I still wonder why Valve didn’t provided Counter-Strike 2 earlier for Linux. The average user of Windows doesn’t know where and who to reports bugs. System tooling for reporting bugs on Windows is also less advanced.

Developers of closed-source applications historically avoided to provide open bug trackers. The literally trained their users not to report bugs. In best case there is a forum. Most of the time only an E-Mail address is available.

View on HN

>System tooling for reporting bugs on Windows is also less advanced.

Windows has plenty of solid diagnostic/post-mortem tools. They also have ETW (Event Tracing for Windows), and they're pioneers in vacuuming up crash reports at scale with invasive telemetry.

If for instance someone tries to run an exploit against you and they cause any instability whatsoever in Windows, the stacktrace can end up on someone's desk at MS and vulnerabilities are regularly discovered and fixed that way.

That doesn't replace a user giving good bug reports, but their tooling is not less advanced at all.

As a salient point of comparison, the most advanced tooling for bug reporting that the Linux kernel has, is "printk a warning into the log and hope someone looks at it"

> the stacktrace can end up on someone's desk

This is chillingly scary and one reason why I use Linux desktop.

Microsoft has good protocols around this to respect people's privacy and devs can only temporarily have access to them.

So devs can see what I've been upto on my own computer.

I assume the problem is the default is "send" rather than "don't send"?

>So devs can see what I've been upto on my own computer.

If you consent to sharing the error with Microsoft. It may contain some information about what was running to assist with finding the issue. There is a strict privacy policy for what this data is allowed to be used for.

>I assume the problem is the default is "send" rather than "don't send"?

No, unless an admin changed the policy.

Ahh, if the default is don't send and it's just an option then I don't see what the problem is

But this is all according to the goodwill of Microsoft, who may very well use it for other purposes for any reason and you'd never know.

>But this is all according to the goodwill of Microsoft

No, there is a privacy policy and they would get in legal trouble if they violated it.

How would anyone know that? Is anyone checking on them? There’s been plenty of instances of companies breaking their own, even legally binding, word

I haven't read the privacy policy (who does?) But I am willing to bet that it doesn't restrict MS to these purposes and in fact allows them to do whatever they want for any reason.

It certainly is. And every time a program is forced to have the crash reporting be opt-in because of the region I live in, I sing Ode to Joy. Don't get me wrong I'm no invasive telemetry enthusiast.

But I wouldn't mind being _able_ to make that choice for some projects I trust. If it were up to me, Linux maintainers would be looking at enough (opted-in) user stacktraces in their code that they'd start thinking of syzbot* fondly.

(* Fuzzer that Linux maintainers love to complain about due to the deluge of rare bugs and crashes it drops on them)

A number of Linux distros and software have automated bug reporting tools, they just tend to require an explicit confirmation before submittingn anything.

E.g. there's Ubuntu's Apport, Fedora's ABRT, KDE's Dr. Konqi.

If anything, tooling for analyzing and fixing bugs (as the developer) on Windows is so much better it's not even comparable.

This is a matter of what you are used to more than anything else. Personally, I'm much happier with debugging tools on Linux.

Reply on news.ycombinator.com