Calstis version 2.8 was installed in the OTFC pipeline on August 30, 2000. Calstis 2.8 is not yet available for download, but will be in a future STSDAS release.
The following OPRs are closed by the release of CALSTIS 2.8:
OPR 42256: CALSTIS4 should not try to allocate zero bytes
OPR 42263: CALSTIS6 must include binning as a factor in flux calibration
Below we give more details on changes that are likely to be noticed by users in their data.
When a STIS NUV exposure is taken with the G230L grating and with one of the wider 52 arcsec slits, the 31X0.05NDC slit can be used for the wavecal. During wavecal processing, calstis uses the occulting bars to determine the offset in the cross dispersion direction. Calstis allocates memory for information regarding each occulting bar, and the out-of-memory condition is checked for by comparing the returned pointer with NULL. 31X0.05NDC has no occulting bars. On a Tru64 machine, the returned pointer is NULL when allocating zero bytes of memory, so it looks to calstis as if the allocation failed, and it stops with an out-of-memory error. This was not seen before because on other machines a valid point is returned, and calstis loops over the number of occulting bars, so the code that would have accessed the memory would not be executed. (OPR 42256)
The 1-D extraction routine in the calstis pipeline software (calsti6), when applied on MAMA hires data, was giving flux values that are a factor two lower than the correct ones. The same applied to CCD data with binning > 1 in the dispersion direction. The output flux for the 1-D extracted spectra were lower by a factor of two or four (depending on whether the BINAXIS1= 2 or 4, respectively) for these cases. This has been fixed in CALSTIS 2.8. (OPR 42263)