This bug was filed from the crash reporting server and is br-f8014787-1f7d-48fb-85f3-6e5e7917b9c5. I can't say what happened. Simple delete or copy. I can't reproduce. http://crashreport.libreoffice.org/stats/signature/CompareSwOutlineNodes::operator%28%29%28SwNode%20*%20const%20&,SwNode%20*%20const%20&%29 There are 49 reports, all Windows. No other bug reported.
Changing it to NEEDINFO until we know the steps to reproduce it. Anyway, thanks for reporting this
It will be hard to reproduce. Since there are those 49 reports, last being 5.3.4.2, I'd rather put this on New. Are you sure Crash Report can't be used without steps? I thought they are not necessary if report available. My report is http://crashreport.libreoffice.org/stats/crash_details/f8014787-1f7d-48fb-85f3-6e5e7917b9c5.
(In reply to Timur from comment #2) > It will be hard to reproduce. Since there are those 49 reports, last being > 5.3.4.2, I'd rather put this on New. > Are you sure Crash Report can't be used without steps? I thought they are > not necessary if report available. > > My report is > http://crashreport.libreoffice.org/stats/crash_details/f8014787-1f7d-48fb- > 85f3-6e5e7917b9c5. I wouldn't use the rule 'crashreport equals to NEW' since even a core developer might not know where the problem is. Besides, having the steps to reproduced easier things for developers and QA.
It's clear steps are wonderful thing but steps are not available here. What I ask is important in QA: if we know that there are no steps, but we have crash report, what then? I disagree with NeedInfo because reporter can't give more. I see 56 reports now for this, so increasing (bad) but still last is 5.3.4.2 (good).
> What I ask is important in QA: if we know that there are no steps, but we > have crash report, what then? I disagree with NeedInfo because reporter > can't give more. What's the point in having it unconfirmed if we can not even triage it? IMHO, the less bugs we have in the unconfirmed list, the more we can focus on those that can be really triaged. Developers can go directly to the crashreport page to see the backtrace in case they need to.
So far we only have one crash in 5.4 branch -> https://crashreport.libreoffice.org/stats/signature/CompareSwOutlineNodes::operator()(SwNode%20*%20const%20&,SwNode%20*%20const%20&) @Timur, Should it be closed while we don't have reproducible steps?
If you deem it so, feel free. But I'd rather have dev take a brief look at the code pointed with this crash. There are 80 reports now and newest version is 1 report for 5.4.1.2.
Fyodor, do you think this one is also related to Bug 94225?
Seems that this bug relates mostly to outlining, as all crushes that I reviewed happen when LO tries to work with SwNodes::m_pOutlineNodes Unfortunately I'm not expert in outlining and cannot help. Unlikely that this bug relates to Bug 94225
Closing as RESOLVED WORKSFORME. No crashes in 5.4.3.2, 5.4.4.2 or 5.3.7.2 -> http://crashreport.libreoffice.org/stats/signature/CompareSwOutlineNodes::operator()(SwNode%20*%20const%20&,SwNode%20*%20const%20&)
There's one in 5.4.3.2 but let's wait and see.