[Development] Qt+KDE collaboration on crash reports

Ville Voutilainen ville.voutilainen at gmail.com
Wed Sep 11 08:29:31 CEST 2024


On Sun, 8 Sept 2024 at 16:27, Nicolas Fella via Development
<development at qt-project.org> wrote:
>
> Hi,
>
> at Akademy I had a chat with Volker Hilsheimer and now want to open this
> to a wider audience.
>
> In KDE we have an automatic crash reporting system based on Sentry.
> Naturally some amount of the crash reports point to the problem being in
> Qt instead of in KDE code. In those cases we of course want to create
> useful Qt bugreports for you. Unfortunately it is not always easy for us
> to narrow down or reproduce the problem so a lot of issues are never
> reported to you, even if they could be potentially interesting to you.
>
> Now I am wondering how we can improve the communication pipeline in that
> regard and make sure you Qt people are aware of the reports we get and
> can give your input on it. I see a few possible ways to go about this,
> which aren't necessarily mutually exclusive:
>
> - We report every Qt-related-looking crash we get as a QTBUG, even if we
> don't have a reproducer or much information on what's going on
>
> - Someone from KDE writes a weekly(?)/monthly(?) summary email with
> relevant looking reports that we would like someone from Qt to look at
>
> - We give some Qt people access to our crash reporting system directly
>
> Let me know what you think about this

I like the sound of the third option, technically, because that should
hopefully contain all the information the dump facility
sent. The first option is perfectly fine if you can include all that
data. Nothing wrong with the second option either.

Does the system send core dumps? What's your general experience with
trying to reproduce the problems based
on the data the dump facility sends?


More information about the Development mailing list