[Users] Meeting Minutes

[Roland Haas]

Present: Josh, Roland, Frank, Erik, Ian, Matt, Peter, Steve, Barry,
Philipp, Bruno

svn to git transition of ET thorns:
* LSU has domain name, has ssl certificates for domain
* LSU will update OS on virtual machines in near future, hold off with
transition until then
* Ian Hinder already has git repositories for the ET thorns which could
serve as the basis for the git repositories
* leave svn running in read only mode afterwards
* git server to use: plain git, gitolite (Ian has experience with that)?
* Open question: few large repositories (or even one) or one repository
per thorn
** no agreement on what to do. Two views are presented to either have
fine granularity (one repo per thorn) or being able to make consistent
changes in several thorns at the same time (few repos)
** Roland and Ian to describe arguments for fine granularity repos, for
experience with large/small repos on the wiki (everyone else of course
is welcome to comment as well)
** question on which (if this is done) thorns to collect in
repositories: by arrangement, by institution, by type of code, how
tightly they are coupled?
** Ian and Barry have been using a git super-repo to track several
individual repos which can be used as a per-user solution for a large
repository, it is however not yet ready for public use
** single repository makes it easy to keep several thorns in sync, and
to get a global overview of what has changed, there are also nicer GUIs
available
** multiple repositories make it easier to pick only bugfixes for a
thorn but but new features from another in the same repository
** Barry mentioned git subtree (https://github.com/apenwarr/git-subtree)
as an alternative for single users
** likely need to modify GetComponents to handle git repositories more
similar like svn

workshop report:
* we had David Rideout gave a talk on how to use Cactus with causal
sets, notes are on the wiki
https://docs.einsteintoolkit.org/et-docs/ET_Workshop_Summer_2013_(Developers_Workshop)
* Ellipitic solver: Eloisa Bentivegna's elliptic solver (multigrid,
works with Carpet grids) is available and described in print and on the wiki
** supports Dirichlet and Robin like boundary conditions
** invite Eloisa to give a ET seminar in one of the next calls
* Chemora (see wiki): meet in Atlanta, wait for confirmation of grant
* ML rewrite: make it easier to rearrange equations so that calculations
can be split if needed to improve performance
** improvement to flow back to Kranc
* Erik reported on new bboxset2 class which improves scaling, paper on wiki
* Erik reported on parameter file generator for benchmarks using
simfactory, Ian added performance regression tests on top of this to Jenkins
* Steve added code coverage tests

MHD:
* Philipp has working cell centred A-field evolution code, working on
cleaning it up

Yours,
Roland













-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 259 bytes
Desc: OpenPGP digital signature
Url : http://lists.einsteintoolkit.org/pipermail/users/attachments/20130807/224d9702/attachment.bin