Hey Robyn,
On Thu, 30 Nov 2006, Robyn Allsman wrote:
Are you saying that you wanted the SVN structure to remain the same as
for DC1? The purpose of the reorg was to optimize the distribution for
pipeline developers - who generally will not want to have all the auxiliary
pipeline software delivered. Also, the package is the granularity
of the pacman/eups versioning.
Here's what I was hearing:
o As far as the deployed structure goes, we have a directory
representing the platform flavor (e.g. Linux). Under that, we have
an "external" directory for 3rd party packages. Also under the
flavor directory are the packages from our LSST code tree.
o Tim offered that we would want some structure for organizing our LSST
packages under the flavor directory, rather than just a flat set of
package directories. I agree, and I don't recall any objection.
o I suggested that we look to the SVN tree for guidance for the
organization.
o I believe it is still a general principle to have the SVN tree
organization reflecting the UML model in a sensible way.
Now this does not necessarily imply that we can't drop a field in our
python package organization. I'm mainly thinking of the SVN and software
stack.
I'm setting up the deployment system to allow one to only install the
portion of the system necessary for pipeline development; however, this
still needs to have the same organization as when the entire system is
installed.
cheers,
Ray
_______________________________________________
LSST-data mailing list
[email protected]
http://www.lsstmail.org/mailman/listinfo/lsst-data
