An input file is constructed using a plain text editor. It must be in linux-format i.e. lines terminated by a line-feed only and not a carriage-return+line-feed as in DOS/Windows.
The input file is parsed using the Parser. The Parser will verify that the commands given in the input file can be executed by the observing program, for example that the frequency chosen can actually be observed by the frontend + backend. When bug-free, the binary input file created by the parser is copied to directory 'schedules'. For more information see the section on parser logic.
A model of scheduling of observing is implicit in the construction of input files using the keywords defined below. All observing is scheduled in the same way, by one scheduler with one input file parser.
To set up observing, the Scheduler menu is used to put the Scheduler into run mode. It then reads the files in directory 'schedules'. The Scheduler memu also has an opion to reload the input files that have been placed in directory 'schedules', if this has changed.
In the NCCS several levels of priority are defined in the scheduler. In order of precedence these are:
Only the last four levels can actually be specified for single-dish observing. VLBI_release is used for observations that are prmitted to run during gaps in a VLBI schedule. These must be agreed to beforehand by the responsible VLBI person, to ensure that the VLBI is not compromised. In general then, only the last three levels are used for normal single-dish observing.
Within each level, prioritization occurs based firstly on whether timed observations are asked for eg at a specified hour angle. These will have a higher priority than untimed observations.
The next sub-level of prioritization is based on the end time of the scan.