Process

[Cees De Groot]

On 12/13/05, Stéphane Ducasse <stephane.ducasse at univ-savoie.fr> wrote:
> I do not understand.
[...]
> When we integrate
>        - once this is working we close by adding with cs includes the issue
>
Well, how do you know which Mantis bugs to close?

I am a member of the Morphic team. The idea is that we go through
Mantis, evaluate bugs, and close them if possible. When I fix a bug, I
commit it to the Morphic repository. Then, after maybe one or two
weeks of this stuff, the Morphic Team Leader decides that there´s a
presentable version of Morphic so it can be submitted to the
inbox--you know, checking that all the bugfixes don´t conflict,
etcetera. So, a version is copied to the inbox which includes some,
say, 20-odd bugfixes and a couple of enhancements the team itself came
up with (obviously, these don´t have a pointer to Mantis'.

Now, because the integrators need to go, for this single MC version,
to all 20 Mantis bugs and press the `close´ button, there´s two things
you can do:
- Say that the Morphic team just copies the last .mcz and let the
integration team walk down the history until they encounter the first
closed bug (which means that you landed at the point of time in
history where you did the closing the previous time);
- Mandate that the Morphic team makes a new version with a comment
collecting all the bug fixes so the integration team needs to look
only at the latest version.

The third alternative, require that every bug fix is copied in a
separate .mcz to the inbox, is clearly not acceptable-it sits too much
in the way of freedom for the stewards team to decide how they work. I
can imagine many very common scenarios where you don´t want to
integrate on the level of single bugfixes...