TABLES 3.0 RELEASE NOTES GENERAL CHANGES A FITS IMAGE extension may be opened as a table in order to access the header keywords. Some applications test whether the input is a table, image or text file by attempting to open the file using tbtopn or immap; these were modified to use either tbtacc or whatfile instead of tbtopn. CHANGES to APPLICATIONS TABLES.FITSIO - CATFITS This task was modified to properly skip the data portion of table extensions with variable length arrays. The section handling the skipping of random group data was also corrected. - STRFITS, STWFITS These tasks will recognize POA extensions as text files that should be converted to FITS tables (and likewise back to text files) to support ST-ECF FOS Post Operational Archive products. TABLES.TTOOLS Existing Tasks: - KEYSELECT Changes were made to free unneeded memory allocated by evexpr. TABPAR A new boolean format parameter was added to allow the user to specify whether the value should be gotten from the table using the print format for that column. The default is yes to preserve the previous behavior in the event it is being relied on. - TCOPY A memory leak was fixed. - TCREATE The limit for input lines was increased from 1024 to 8192. - THEDIT An example was added to the help file illustrating the use of the task on an image extension header. Changes were made to free unneeded memory allocated by evexpr. - TINFO Image extensions now are described as "image" instead of "primary header" for the value of the 'subtype' task output parameter (now that image extensions are permitted as input). - TJOIN If only one tolerance value was specified when multiple join columns were specified (so the same value was to be applied to all columns), the amount of space allocated for the tolerance array was only large enough for one value. Space is now allocated for a value for each column. - TPRINT A specific format is now used for int and short columns. This change was made because if a print format was x or o, the printed value would be hexadecimal or octal respectively, but it could be misinterpreted by the user as decimal. - TREBIN The task was failing to properly delete the output FITS table when the independent variable in the input table is not monotonic. This resulted in a confusing error message from tbdel instead of one regarding the input variable. This has been fixed. - TSELECT tblsearch.x now releases memory allocated by evexpr. SYSTEM LIBRARY CHANGES - CFITSIO Upgraded to version 2.401 The default number of HDU's that can be handled was increased from the CFITSIO install default of 1000 to 15,000. - GFLIB Previous enhancements to GFLIB allow opening FITS files mapped from GEIS files (where groups appear as image extensions). There was a bug that caused history records to disappear in propagated headers. This was fixed by keeping track of history records in a temporary spool file. The incorrect FITS extension spec was being built when the file name was of the form [extname, extver]. This was fixed by appending the ksection string provided by imparse. Unused function declarations were removed. MAXDB was changed from 64 to 128 to allow opening more FITS files. - TBTABLES A FITS IMAGE extension may be opened as a table in order to access the header keywords. Some applications test whether the input is a table, image or text file by attempting to open the file using tbtopn or immap; these were modified to use either tbtacc or whatfile instead of tbtopn. Newlines are now removed from error messages that are passed as the second argument to the error function. (tbtopn.x) The file name was added to the error message for the case where the table file does not exist or cannot be opened. The buffer size for the error message was increased in the event the file name is long. (whatfile.x) Now uses tbtacc instead of tbtopn to test for a table. (tbtacc.x) For a FITS extension (or primary header), return NO if the subtype is image. (tbfhp.x) When keywords are blank, sprintf only eight spaces before the value to agree with CFITIO behavior. Previously copying header records with blank keywords resulted in the value (comment) being shifted two spaces to the right on a copy.