S. Digel, 6 March 2005
This is fine with me. If it is greater than ~1 week, we might want to try putting it in the GSSC's server; FT1 files run about 50 Mbyte/week (depending on what lower energy limit is adopted).
What about simulated GRBs, and GBM data products for them? I do not know yet whether the GRBs can be defined in the same model as the other sources. My understanding is that Nicola's GRB source can write an ASCII file for each GRB that it generates that has the information that David's software needs to generate equivalent GBM data products.
Enough has changed since the first checkout, e.g., regarding using DSS keywords instead of ROI files, that an updated tutorial on making a likelihood analysis would be useful. There's still some confusion in my mind about exposure calculation tools - which tool does what and what needs to be run before a likelihood analysis.
I think we need a tutorial, too, about using the likelihood classes from within Python.
The Python GUI interface to likelihood and for defining source models is available. (At SLAC, I think that the latter probably still works only on noric13.)
I need to try out the MapCube version of the interstellar emission model (generated by GALPROP) as a source for likelihood. I think that it will work, but I am curious about performance relative to a Map source.
A question about an 0.5 day difference in reference dates between the astro classes and glbary still needs to be resolved. That, and the difference in units for spacecraft position between the FT2 file (km) and what glbary expects (m) may prevent including glbary in checkout 2. (The HEASARC convention is what glbary expects - and we will convert to using m for distances.)
I am not sure what is new to be tested, other than new binning schemes in evtbin. The Band function 'source' for likelihood analysis has not been thoroughly exercised yet.
Will we have the ROOT GUI interface to the parameter files? Is that interface something that sits in front of the files or would it require code changes in the tools?
How about Python interface to parameter files? It is available, and I would not discourage people from using it.
Regarding the installer, my understanding is that we won't have a Windows distribution for checkout 2 either. My impression is that we are at least somewhat closer to having the Release Manager routinely attempt builds and tests under Windows, but even a manual build doesn't work right now. See also Likelihood build results from Jim.
Under Linux, there may be an problem with where the installer has the tools look for their pfiles by default.
JIRA issues need to be closed so we don't raise the same ones again in checkout 2.
Some useful utilities don't exist yet - e.g., for displaying response functions, maps, source models, astronomical catalog overlays - and won't exist for the checkout.
March 14-April 4 - ok? Is there enough that needs to be resolved yet that we should push this back another week or two?
I've contacted the reviewers from the first science tools checkout and all but one who I've heard from so far are willing to be reviewers again.