[kubuntu-devel] Muon Discover

[Harald Sitter]

On Fri, Jan 10, 2014 at 11:52 PM, Valorie Zimmerman
<valorie.zimmerman at gmail.com> wrote:
> Thanks for your gentle reminders to file bugs, Harald.
>
> On Fri, Jan 10, 2014 at 9:06 AM, Harald Sitter <apachelogger at ubuntu.com> wrote:
>> On Fri, Jan 10, 2014 at 5:58 PM, Volkan Gezer <volkangezer at gmail.com> wrote:
>>>>>
>>>>
>>>>> Mature enough? It does not seem so to me.
>>>>
>>>
>>> +1
>>>
>>> Some problems I experience with Discover:
>>>
>>> * Closing it before POPcon result are shown in the main window causes
>>> sigfault sometimes.
>>> * Using back arrow after making a search either freezes or displays
>>> all installed applications.
>>> * Ratings does not work. (we cannot rate etc)
>>>
>>>
>>> My question is why did we remove Muon Installer? It was ok IMO.
>>
>> Because apparently discover only got buggy in the last 48 hours. Very curious.
>
> However, I see your comments and feedback on
> https://bugs.kde.org/show_bug.cgi?id=329800 and that seems the bug
> that bites me.
>
> Searching bugs.kde.org for "muon discover" reveals lots of crashes and freezes.

All software has bugs, and the volume of reports for discover in
particular is not actually overwhelming. So since the only hard metric
we have right now (to be changed starting with 14.04) are bug reports
I have to say that all the data suggests discover has at least
adequate quality. That is not to say that it did not have show
stoppers, but I think just about all of them have been resolved in
like a week after reporting. And that is the crucial thing really,
unless crashes and the likes are vigorously reported as bugs it is
nigh impossible to label a defect as high impact and thus worthy of
asap handling.

That sounds a bit silly but bear in mind that we are working with
limited resources (<=24h/person/day) and therefore I personally do not
consider it practical or efficient to have developers treat every
crash as super important or troll the internet to find out whether a
crash/bug has gotten sufficient whining on forums etc. to be
considered for immediate fixing. So at least I hold on to the metrics
I've got, which are bug reports and the amount of attention a bug gets
(amount of incoming duplicates, amount of comments and so forth). If
there is no indication that a bug will/does impact a sizable number of
people or has other factors that would qualify it as important, it
does not get the attention that it needs which results in unhappy
people and unhappy comments about the software and finally in unhappy
developers because people do not like their software.

It's a poison trap that can easily be averted by reporting bugs and in
general constructive discourse about issues. Every KDE application has
a built-in functionality to do just that; in the menubar -> Help ->
Report Bug... but if it is not used, all is vain.

Why do the discover comments bug me so much you may ask yourself and
the answer is very simply that it most certainly has not suddenly
gotten quality issues in the past week. They have been there all
along, and no one bothered to do anything about them... and yes,
reporting a bug is very much doing something about it.

What transpired was: A transition from muon-installer to muon-discover
was publically targeted for 13.10 [1]. In fact I explicitly pushed for
early implementation of the necessary changes to have enough time for
testing *before* feature freeze so we could have reverted back to
muon-installer well before the final release of 13.10. Due to lack of
quality control measures (being addressed progressively since then)
and lack of actual feedback of any kind before the final, less than
adequate quality was released and many tears were shed once we
noticed. We then fired numerous updates to resolve the situation
[2][3][4].

This should not have happened. It just shouldn't have. There was no
reason for this to happen. Early september JR sent a mail about the
muon quality saying that all was nice and dandy - no one raised
concerns. And in november, *after* release, people then start to
complain? Why? I absolutely do not understand this. If people do not
feel the need to do pre-release testing, then we might as well stop
doing the pre-release nonsense, and instead spend our last bit of
remaining sanity on manually executing tedious test plans to ensure
the software we ship is actually release quality.

Usually I would not care about this post-release-issue-creep-nonsense
(as can be observed on web forums every time after a final release has
come out, as if we had developed the thing in secret and now everyone
is surprised that there are issues that could not possibly have been
catched before release). But when it happens on the mailing list that
calls itself kubuntu-devel I do take personal offense. I do believe
everyone here has an interest in producing a high quality product and
to that end should do what needs to be done (e.g. take the pre-release
for a test drive and report bugs at least once in the 6 months we
spend preparing a new version). That did not happen so now you get
deal with a sad pandalogger.

tldr: always report bugs; for everything; if the reports don't get
attention: complain on this here mailing list; if complaining doesn't
help: tell apachelogger.

[1] https://trello.com/c/DplmVapI
[2] https://trello.com/c/CSivV2uq
[3] https://trello.com/c/W2OQ339e
[4] https://trello.com/c/EUHfjxSC

HS