Standard sequence: save data in a scratch file

Purpose: dispose of an observation's file for later use in calibration procedures.

Typical course of events:

• System renames the observation file (a FITS file) according to the given scratch-file number. The final file has the name sn.fit where n is the scratch-file number. Lower case is used and the number is not padded to a standard length with leading zeros.
• If a scratch file of the given name already exists, it is lost and the new file takes its place.
• System puts the final file in a directory named for the the date in the format yyyymmdd, e.g. 19980816 for 16th August 1998. The date for the directory is the UT date at the start of the night; it changes at noon.
• The system puts the directory in the general partition /obsdata on the disk server. This one partition is all the raw-observation space the telescope is allowed and is assumed to be large enough to hold at least one night's observations.

Variations:

• The big partition /obsdata on the disk server is not available or is full up with old observations. The system uses an alternative set of disks. Performance may be reduced, but the system is not rendered useless or unreliable.
• The observation produced more than one file because there was more than one readout point. The system assigns scratch-file numbers starting with the given number and counting up. The numbers map to readout channels in a standard way that is recorded in the FITS headers of each file.