===================================================================== Comb News and Views ===================================================================== ------------------------------- Changes for Comb 2000.0 ------------------------------- 11. String lengths increased for BUNIT, TELESCOP, OBJECT Formerly these FITS header variables were restricted to 8 characters. That restriction is lifted. They can now be up to 68 characters (as can any FITS string). 10. Dates now stored with stacks. Previously, only scans had dates associated with them. With "st", the date was lost. Now, the date of a scan is stored with the stack in the j2sec variable (see news item #9). The behavior of pd date: is thus changed to print out the date in FITS format (CCYY-MM-DDThh:mm:ss) of stack 1. For stacks created in pre-2000.0 versions of comb, there will be no date stored with the stack (.j2sec=0), so no date is printed out. This same date string is written out by "wf st:" in the DATE-OBS keyword. The creation date of the file is written in the DATE keyword. Similarly, a FITS spectrum with DATE-OBS defined has the equivalent value stored in j2sec. All output dates are UT. Input dates are assumed to be UT. 9. New stack header variables j2sec and ftrm. The number of seconds before or since J2000 is now stored in .j2sec. Negative values of j2sec mean before J2000, positive mean after J2000. The value of j2sec is converted to calendar time for spectral plots (pl). Zero (2000-Jan-01 00:00:00UT) is ignored since previous versions of the stack header have this location zeroed out. So if you happen to take data at exactly 2000-Jan-01 00:00:00UT it won't get displayed! j2sec is set either from the J2Second header variable (for Bell Labs format), from the Class3.ut variable for NRAO format, or from the FITS DATE-OBS keyword for singledish FITS format. For co-added stacks (stacks with multiple scans), j2sec has the value of that of the first scan in the list. If Fourier components are removed from a stack (see ft ?! and News item 3 below), the number of components removed is stored in .ftrm and display in a spectral plot next to the baseline order. These two variables took the part of the "spare" header variable isdum, which now down to the length of 2 short integers. 8. Baseline order FITS keyword for stacks The baseline order removed from a stack is now written out in FITS spectra (wf st:), using the keyword BASELN. A value of -1 means no baseline has been removed. 7. New vertical offset option to "pl" The plot command pl now has the option voff: to add an arbitrary vertical offset to the current pl. This is useful for overlaying spectra without having them right on top of each other. E.g -> rt 10;pl; rt 20; pl ovl:voff:3;rt 30; pl ovl:voff:-3 would plot stacks 20 and 30 over stack 10 with a vertical offsets of 3 and -3, respectively, in the current plot units. Note the offset can be positive or negative and reverts to zero if not specified. 6. NRAO SDD and PDFL scans can now be read on little-endian machines (e.g, PCs). At long last, Marc has made the NRAO scan routines working on PCs (under Linux at least). 5. Comb is Y2K compliant! Actually there were only a few minor issues with Comb and the Year 2000. The main one is that COMB now writes FITS files with the DATE value according to the new FITS standard: CCYY-MM-DDThh:mm:ss in UT. 4. New predefined constants For your convenience, AMU, C, E, G, K, MSUN, PC, and PI are now predefined. You can access these as normal variables, eg. -> p .PI 3.14159 For the exact definitions, see dv ?!. 3. Upgraded Fourier Transform (ft) command The Fourier transform command ft has been upgraded by Tony Stark. New features include removal of Fourier components and number of components to fit. (see also news item 9). 2. Stacks format change The maximum number of channels allowed in comb is now 16224, down from 16384 in previous versions. This change was required because the length of a stack index is limited to be a unsigned short integer and must be a multiple of 32 for use arrays to work as defined. 1. 'up' (update search directory) and 'ns' (name stacks) upgraded This was to fix a problem related to moving stacks files between machines with different byte orders in their words. (e.g. a PC and a Sparcstation). When opening stacks directories the files are now checked for the correct byte order in a way which should work except for some very unlikely cases. 'Ns' checks that the order is correct and 'up' checks that it is wrong. In either case if any errors are found, the number of errors is printed as well as the number of errors which would be found if the file were "up swap:"ed and the user is asked to proceed. In addition, 'up' has a new option, tbo:, to just test the byte order and return. The same error report described above is given, but no bytes are actually swapped. Use this if you aren't sure about the origin of your stacks. The documentation for 'up' ('up ?!') has been almost completely rewritten. The old version was useless and the new should be much more helpful.