Earlier this day my PC froze to the point again that I had to sigterm all but init.
Checking my logs and trying to figure out what the bug is was so annoying that I was like:
Hey, why doesn’t a machine do this for me?
Hear me out:
We could centrally collect the logs (basically ALL logs, like, I don’t know, kernel logs, qt logs, journal logs, etc.) in one central place, e.g. here on KDE, and let an algorithm (a learning algorithm) find the issue for the crash by finding patterns in crash reports.
Also, automated reports would not require any user interaction. (I HATE “Send report?” buttons)
Now, since a lot of bugs are actually already sort of known and reported somewhere, this its self would not be all that interesting, BUT we could EASILY assign PRIORITIES to bugs depending on how often they are reported in crashes!!
But what about privacy?
Well, we could either never let the logs go public or we could try to check the logs for personal data like IP addresses and user names. Also, we could just show a little warning the first time there is a bug reported, which defaults to “Don’t ask again”.
The current state of …
- Ok my pc/ program crashed
- Ok a quick search did not find a similar report
- Ok let’s take the logs and open a new issue somewhere
… is absolutely not acceptable, if you ask me.
I hope somebody actually considers this, as I think if properly done we could actually get proper bug tracking, which is separating bug reports from crash reports. (E.g. the bug gets detected on the central log collecting page, and then somebody condenses the info from multiple crash logs to a single bug report for e.g. gitlab in one go)