[Libreoffice-qa] Minutes - QA related - TSC call 2012-01-26

[Alex Thurgood]

Le 28/01/2012 13:36, Pedro a écrit :

Hi Pedro,

> My plain opinion about this: all developers should focus on fixing
> regressions. And then on fixing bugs. And only after that in adding new
> features/bugs.

I would tend to agree, but it is true that I'm not a developer, and thus 
my motivations lie elsewhere. I can not speak for the motivations of any 
given developer within this project.

>
> Simply put: Regressions + Bugs cause to LOOSE existing users.

Absolutely.


>
> If someone upgrades to get rid of a bug that was present in his current
> install and finds that it is fixed but other features that he needed stopped
> working, what action should he take? Go back? Wait?

My personal approach is to stick with the old. That's why I still have 
LO 3.3.4, and even OOo 3.2.1 - why change for some allegedly 
indispensable feature when the stuff you are used to working with no 
longer does ? That also goes for any general recommendations I might 
make about the software.

>
> Most users that try a free software if they find an obstacle they simply
> uninstall it. Very few will bother to find a solution. Even less will bother
> to report it. Only a rare minority of these will actually register in the
> tracker and actually report it.

Agreed.

>
> How to improve this? Eliminate all bugs *. Eliminate barriers to reporting
> bugs.
>
> * I know this is an utopia. But you can still set it as a target.

Also agreed. Where I disagree with the current approach of the project 
is that it is expected that the masses will test the latest and greatest 
dev release and report the bugs they find. IMO, this is not only 
unrealistic, but also unworkable, and it shows in the final product 
because the point zero releases up til now are still full of 
regressions/bugs that only get fixed 3 or 4 releases further down the 
release line (if at all). I feel that this will only likely change, if 
and when the project has enough developers to be able to "assign" 
(either voluntarily or through coordinated agreement) a group of them to 
fixing regressions and bugs, which (I would assume) is a rather 
ungratifying task for most of them. Most potential users are as we say 
in French "attentiste", they wait for a finalised product to be released 
before taking the plunge. Why risk installing something that might 
corrupt your files, and lose the work you have saved over the past 3 or 
more years ? This is beyond many people's comprehension, and certainly 
in the real business world, where most of the regressions tend to get 
discovered. How many IT support managers would risk their necks and 
their jobs for that ? Freedom has a price, but most people are 
unprepared to pay that price, be it through selflessness or even just 
personal survival.

I still think that paying for bugfixes is a good idea, with an 
expectation that they will be fixed within a given timeframe. I also 
know that this idea is not particularly welcome within this project, so 
I'll keep on testing the new releases, as and when I've got the time, 
but recommend the older 3.3.x releases, or even old OOo 3.2.1 for most 
daily work.


Alex