[Nemiver-list] nemiver release scheme starting from the 0.5.0 version

[Dodji <dodji domain hid>]

Hello,

As you might have noted, since the 0.5.0 opus of nemiver, we tried to
do more frequent releases. As a result 0.5.1 and 0.5.2 got pushed out
at a faster pace.

I think it would be a good goal to try and push out at least one
bugfix or translation update related release per month.
Each bugfix/translation release will see its micro version number
increased,  e.g: 0.5.1, 0.5.2, 0.5.3,  etc.
New feature releases will see their minor version number increased,
eg: 0.6.0, 0.7.0, 0.8.0 etc.

Bugfix only commits are pushed to trunk, in svn. I will then take care
to cherrypick those bugfix only commits into the current stable branch
(or instance the current stable branch before 0.6.0 gets released is
the 0.5 branch, browsable at
http://svn.gnome.org/viewvc/nemiver/branches/0.5/) and manage to push
out a release each month, based on the content of that stable branch.

In the mean time, features can also be delivered to trunk.

For this to work, I think it is _very_ important to follow at least
two guidelines:

1/make sure to follow the coding style of the project very very
closely so that I don't  have to reformat the source code to comply
with them. Those coding guidelines are browsable at
http://svn.gnome.org/viewvc/nemiver/trunk/coding-style.txt?view=markup.
This is important to follow because otherwise, reformatting could make
merging unessarily complicated for me. Until now, I have been quite
relax on those coding guidelines. Usually when I see a piece of code
that was not compliant, I just change it to make it comply. But now I
think we should make an effort to be way more strict in that regard to
get the thing right in the first place. Lazyness should not be an
excuse anymore :-)

2/keep up with the informa one-commit-equals-one-bugfix scheme that we
follow today already. That will ease merging the bugfixes in the
stable branch.

I am not too much for the stable/unstable release scheme. To me, a
release should be as much useable as possible. We tried to be paranoid
and defensive in the way we code to ensure that we have as less
crashes as possible. Furthermore I believe our automatic  regression
test suite can help us catch a fairly significant number of
regressions before we release. As a matter of fact I _think_ we can
just keep on pushing new feature releases out as we have been doing
already without being worried too much about pushing "unstable"
releases out first.

With all this on, I hope we will slowly catch the GNOME release
schedule train and follow it. I believe this will be a good exercise
for us and will lead to a better experience for our users eventually.

Thank you for reading so far.

-- 
Dodji Seketeli
http://www.seketeli.org/dodji