[Libreoffice-qa] End of the line for 3.3 family and regressions

[Petr Mladek]

Pedro Lino píše v Pá 09. 12. 2011 v 20:34 +0000:
> Hi all
> 
> Looking at the Release Plan chart
> http://tdfsc.files.wordpress.com/2011/07/libreoffice-versions.png
> and wiki
> http://wiki.documentfoundation.org/ReleasePlan
> I guess version 3.3.4 is the end of the line for family 3.3.

IMHO, community can't support too many stable versions. Customers who
need it, have to pay someone for it. We should leave 3.3 and concentrate
on making 3.5 the best release ever.


> This means that for many users (and especially for companies, which
> only use the most stable build) the next version will (sometimes
> forcefully :) ) be 3.4.5.
>
> I think this carries (from a QA point of vue) a much heavier
> responsibility and care than the change from 3.4.4 to 3.5.0 (which is
> "experimental")

I think that there is no need for panic :-) 3.4 has been out since, Jun
1, 2011. We did 4 bug fix releases. There was plenty of time to escalate
bugs that would break people from using this version and there were many
possibilities to get these fixes.

In addition, as Michael said, 3.4.5 need not be the last bugfix release
for the 3.4 version. I am sure that we will do another one if there is a
reasonable demand from users and fixes from developers.


> The planned release date for 3.4.5 is on January 11, 2012 and
> apparently there won't be any Beta releases, it will jump directly to
> RC1.

We should not need beta versions. We do not do big or dangerous changes
between bug fix releases. All changes are reviewed. If we would want to
include anything dangerous, we would ask for extra testing. I am sure
that we added regressions between bugfix release in the past (we are
just humans) but I do not remember anything serious.


> I urge everybody to make sure that EVERY regression detected from
> 3.3.x to 3.4.x is fixed/added to the 3.4 branch

This is not realistic. Every application has bugs and every version adds
regressions. This is true for any software, not only for LibreOffice. It
would take thousands man-years to fix all bugs in LO. If you want to
release software at some point, you need to do compromise. There are
good processes (severity, priority, most annoying meta bug) that helps
to focus on important things, optimize the cost and value, and make this
working in a real life.


Best Regards,
Petr