On Oct 14, 7:28 am, Rob Beezer <[email protected]> wrote:
> I wonder if the "lieutenant" model used by Linux kernel development
> might be helpful here?  

Ok, I've read this and to make sure you all are not talking about
different things: there are two ways to split the workload:

- horizontally: Sage itself is split into modules (like the notebook
gets its own project status). that's similar to linux in some way,
where a sub project is only concerned for a specific part of linux and
ignores everything else. At the center of the storm, a central release
manager just takes a new version of a module and tries if it works --
on the other side, satellite projects know about his current status
and can use his one to make their modules work with his main project.

- vertically: the whole live cycle of a patch up into the finished
release is split among different persons.

i think, horizontally is good. The total amount of code lines shink,
less to worry about, more specialized knowledge (i.e. a module is only
about combinatorics), etc.

vertically bad, because of the communication overhead. We can work
around this with "buffers", collecting patches and so on, but in the
end two different people are working against each other, dependencies
break all the time and one says a patch is ok, the other one sees that
the patch makes troubles and pushes it back. I guess we would need a
huge set of rules...

H
--~--~---------~--~----~------------~-------~--~----~
To post to this group, send an email to [email protected]
To unsubscribe from this group, send an email to 
[email protected]
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
-~----------~----~----~----~------~----~------~--~---

Reply via email to