Updating_Cdbs

Rationale

This document was written to assist anyone (including myself) who may need to update the cdbs tasks to support new reference files or instruments. Since the calibration pipelines are continually being updated, there is a need to also update the cdbs tasks to track these changes. Although I have tried to make these tasks as independent of instrument specific information as possible, that is not completely possible. This document only describes changes to the cdbs tasks, it does not describe the changes to the cdbs database which are also necessary. I will continue to update this document as I discover more dependencies on instrument specific information.

Key assumptions

When the new version of the cdbs tasks were written, certain key assumptions were made in its design. Violating these assumptions may require rewriting significant of the cdbs code and I hope this can be avoided. The current cdbs code assumes that:

• CDBS processes three kinds of files: images, tables and loadfiles. Images and tables are whatever can be accesed through the corresponding iraf interface. Load files are created by the cdbs tasks to contain information to be loaded into the database. Other types of files and fits file containing a mix of tables and images are not supported.

• The type of a file can be determined from its extension and from information contained in the header. In the case of fits files, this information must be in the primary header. The type includes both the data file type (image, table, or loadfile) and reference file type (such as flat field or dark image.)

• Reference files are segregated by instrument and data file type. Thus there are two directories for each instrument. The logical name is constructed from the instrument code followed by "ref" if the file is an image or "tab" if it is a table.

Adding Reference Files

If a new reference file is added to cdbs, the data files in the data subdirectory must be modified. A new entry must be made in the file cdbscatalog.dat for both the reference file and its associated loadfile. And template files must be contructed for the reference file and its associated loadfile. This change to cdbscatalog.dat is necessary so that the cdbs tasks can properly identify the file and the template files are necessary so that certify can be run on the new files.

Adding a New Instrument

If a new instrument is added, additional changes are needed to cdbs beyond the changes necessary for any of its new reference files.

• The macros in config.h which identify the instuments and their letter codes must be updated.

• Getuniquname.c in uniqname determines the method of creating the unique name from the instrument, so a new case must be added to the select block that chooses the method.

• The data file directory.dat maps iraf logical directories to host os directories. New rows for the directories containing the reference files and throughput component tables associated with this instrument must be added.

• The data file thrucat.dat maps throughput table names to the logical directory name containg the component throughput tables. A new entry is necessary for the new instrument.

• Several arrays in global.h in the getref source code must be updated with information about the new instrument. These arrays contain the instrument name, the cdbs file and row level table names, the dads bestref table name, and the field in the bestref table containing the observation set name.