STScI-Citizen-Science / MTPipeline

Pipeline to produce CR rejected, astrodrizzled, png's of HST WFPC2 solar system data.
6 stars 1 forks source link

Revise settings.yaml #130

Closed ktfhale closed 10 years ago

ktfhale commented 10 years ago

settings.yaml should be edited to only have the variables we actually want to use, and the rest of the pipeline should be edited to actually use those variables.

We should have run_imaging_pipeline use the number of cores specified in the settings.yaml file. At the moment, it's hardcoded to use the total number of available cores, minus 1. At the moment, I've hardcoded the number of cores to 4 in master, for ease in running the pipeline.

We should delete the fits file input/output paths for the various instruments. Inputs are currently set by the -filelist input argument when calling run_imaging_pipeline from the terminal, and the pipeline distinguishes instruments using header data anyway. While I don't know exactly how it works, pipeline outputs always seem to go in the same directory as their respective input _c0m or _flt file. This may not be desirable behavior for the png outputs.

We'll keep the database and email flags, as well as the logging path. But I've noticed that actually running the pipeline involves creating a run_imaging_pipeline/ directory inside whatever folder you specify as the logging directory. I'll fix that too.

ktfhale commented 10 years ago

I've updated run_imaging_pipeline() to use the num_cores entry from the SETTINGS dictionary. I've also cleaned up the template_settings.yaml file, although I've kept the old one around for now as old_template_settings.yaml. Finally, I've gotten rid of module_name from the logging path, so that the pipeline's logs actually go into the path specified in settings.yaml.

If we do implement logging for the other modules, we could put the module name back in the path. But at the moment, the pipeline fails to run if the user doesn't beforehand manually create a run_imaging_pipeline directory inside the directory specified by logging_path, which they had no way of knowing they needed to do. If we want to add module-specific logging, we should at least have the pipeline automatically create those folders.

I think this branch is ready to be merged.

ktfhale commented 10 years ago

I added module-specific logging back in, and now have the logging path and directories being automatically created (with their permissions set) if they are not already present.

ktfhale commented 10 years ago

Now that the branch has been merged into master, I think this ticket can be closed.