[Bugsquad] Bugday postponed, looking for new target [Was: Krita Bugday 1]

[Frank Reininghaus]

Hi,

I've CC'd Peter so he can comment on your suggestion. I'll add some
comments below...

2010/7/20 Michael Leupold <lemma at confuego.org>:
> Michael Leupold wrote:
>>>    Alternatives:
>>>
>>> * Make it a Crush day (search for undiscovered bugs).
>>> * Try the irreductible Konqueror. (>3000 open bugs)
>>> * Try other Koffice apps (kword, kspread, kpresenter, kexi)
>>
>> Krush might be good for KMail if it's worth krushing yet. Alternatively
>> I'll be asking khtml developers if there's a dedicated batch we could
>> triage - probably better than assorted Konqueror bugs.
>
> I talked to Konqueror developers and it doesn't seem like there's an easy
> target that would be suited to novice triagers. However I think dolphin bugs
> might be nice. They seem to be quite numerous recently, even regarding
> crashes.
>
> Advantages:
> - well-defined area of triage (single application with single purpose)
> - should be mostly easy to understand (no html or similar)
> - plenty of bugs hopefully means plenty of duplicates :)

My comments:

1. There have been lots of crash reports which are somehow related to
Nepomuk/Soprano recently. I've gathered some information about these
at

http://techbase.kde.org/User:Frank78ac/SopranoCrash

I think all reports listed on that page are due to the same bug, but
I've been rather conservative concerning duplicate-marking so far,
i.e., I've only marked reports with mostly identical backtraces. There
are still lots of reports which are not listed on that page and which
should probably be marked as duplicates of one of these reports, but
the problem is the backtraces are usually not equal. A common pattern
is that two or more threads try a malloc/free simultaneously and one
of them asserts, which indicates that there is some kind of memory
corruption.

Triaging these crashes is not suitable for newcomers IMHO.

2. There are some more crashes which get reported repeatedly and which
are more easily marked as duplicates, but they could be hard to find
in the big mess of the reports I've mentioned above.

3. Dolphin's non-crash reports do not have many duplicates, I think.

4. One idea that I've had recently and which could be good for
newcomers: Go through all normal and minor Dolphin reports (i.e., no
crashes, no wishes), check if they can be reproduced and add the
"reproduceable" keyword to these reports. This could have two
benefits: Make it easier for developers to find good bug-fixing
targets, and provide a place to get started for potential new
developers. In a second step, one could go through the "reproduceable"
bugs and check which would be good "junior jobs".

Cheers
Frank