NUCOMP 2009-9-9

Detector simulation was mentioned - do we care about this? if so, at what level except running it?

offline software tools discussion

Looks like a list of products that the presenter likes or is familiar with.

general questions -

  • how do you make decisions about these products? * how do you judge what a good and bad product are? * what is the overall goal here? * who will maintain this software and integration over the experiment?

CMT questions and comments:

  • have you configured it? * have you changed or added an external product in it * all the users look like users of GAUDI - who uses that is not in this category? * is it easy to pull GAUDI down and build it with CMT?

version control - what is the criteria for choosing one or the other? what are the use patterns?

web-interface - what is the purpose of this tool?

issue tracker - only trac mentioned.

interesting - dayabay has build and test stats on the web - using Bitten python module?

novainst - automatic build system for external and internal packages (offline)

looks like NOVA offline has decided on this tool set - is this true? (no, and not for the whole experiment?)

looks like the decisions are based on what GAUDI uses.

Heidi - says the requirement is that "we are using this"

Marc Mengel's talk

looked at CERNs use of repositories. some are using git there.

Robert makes a point that multiple SCCS means more difficulties in experiment support.

Adam makes a point: JIRA is used for [operational] incident tracking, and not really for software problems and features.

Heidi puts constraints on external products that she does not seem to put on the product that they picked to use.
What about CMT? Is it subject to the same constraints?

I still do not understand the release and build situation. One problem is porting the FMWK to other build systems and not
thinking of it as an external product.


talked with Rob, Ron, and Marc P. about the above session for another hour at the end of the day.
What was called out here was the differences in project management between SVN and git. The biggest
is that SVN can still be viewed as a tool that you place your entire project into one repository. The
git tool is not like this. We would need to generate many small repositories and have a release tracking
database to know the relationships between all the repositories. This system or database can be viewed
as a repository dependency tracking database.