A Query on a sudden increase in the bug dependency of LibreOffice

[Ilmari Lauhakangas]

On 26.4.2021 22.16, Hadi Jahanshahi wrote:
> Dear developers,
> 
> I have a question regarding bug dependency in LibreOffice.
> While I was studying the blocking and blocked bugs, I found that in 
> LibreOffice, after 2017, there is a sudden spike in the number of the 
> found dependencies [1 <https://arxiv.org/pdf/2011.05382.pdf>].
> image.png
> /Figure1- The monthly evolution of mean depth and degree of the Bug 
> Dependency Graph for the LibreOffice project (x-axis corresponds to the 
> year and /the /y-axis corresponds to the mean depth and degree of the 
> graph)./
> 
> May I ask what is the interpretation for that sudden increase? What 
> happened that developers started to find these many bug dependencies?
> Thank you for your help in advance. I hope that I contacted the correct 
> list.

Hello,

others have already explained that your chart ends up mainly looking at 
the categorisation of our bug reports. You should be aware that Mozilla 
also uses meta bugs as a way of tracking bigger tasks. Random Firefox 
example, about adding VA-API hardware decoding support on Linux (you 
need to be logged in to view the tree): 
https://bugzilla.mozilla.org/showdependencytree.cgi?id=1210727&hide_resolved=1

So meta bugs are basically just another way to associate our reports 
with useful meta data. Many of them could be replaced by adding another 
level into our product - component hierarchy.

You can study all of our meta bugs conveniently here: 
https://wiki.documentfoundation.org/QA/Tracking_Bugs
As you can see, some of them are meant to be permanently open and some 
of them have a clear scope (such as fixing all regressions from some 
particular change).

It is good that your paper is still a preprint because it needs 
revision. There are several parts that do not make sense to me. For 
example, on page 15 you say

"The sudden change in decision of triagers for not prioritizing bugs 
with high depth leads to accumulation of the bugs in the system and a 
much higher complexity of the BDG."

This conclusion has nothing to do with the reality of LibreOffice 
development, because the cases where prioritising in itself leads to 
action are not very common. Typically the reports of highest priority 
and critical severity are brought to the weekly engineering steering 
committee meetings for discussion.

If you want to discuss more, we can schedule a call. You can email me 
directly.

Regards,
Ilmari Lauhakangas
Development Marketing at The Document Foundation