Open mathiaswagner opened 9 years ago
Hi Mathias,
It would be good to think about an interface similar to the one used by QOPQDP, which has a pretty flexible system for setting options.
Best, Carleton
On 4/9/2015 12:54 PM, Mathias Wagner wrote:
A lot of things in the MILC interface are currently hard coded. We should discuss whether we want to expose some of them to the user or to the MILC code. One particular example is the reconstruction of links.
We can certainly expose these options in the interface but should coordinate with MILC where the user should select these options. (QUDA make.inc, MILC Makefile, ...). Just moving from having them hard coded in QUDA to having them hard coded in MILC is probably not very useful.
— Reply to this email directly or view it on GitHub https://github.com/lattice/quda/issues/220.
Carleton DeTar Department of Physics and Astronomy University of Utah
Thanks for pointing that out. I will look at the QDP interface and how that can inspire an update QUDA MILC interface.
I guess we don't want to break the QUDA 0.7 / MILC 7.7.12 compatibility for now.
When we do work on the interface we should also check for hard coded verbosity.
I suspect this is not something that we can do for 0.8, since this would be a disruptive change. Marking as 0.9.
A lot of things in the MILC interface are currently hard coded. We should discuss whether we want to expose some of them to the user or to the MILC code. One particular example is the reconstruction of links.
We can certainly expose these options in the interface but should coordinate with MILC where the user should select these options. (QUDA make.inc, MILC Makefile, ...). Just moving from having them hard coded in QUDA to having them hard coded in MILC is probably not very useful.