Starlink Software Collection - Lehuakona Release Notes
All documentation has been updated to use a current version of latex2html. Most now use LaTeX2e too.
- This will be the last supported OS X PPC release.
- This will be the last supported OS X 10.4 (Tiger) release for both PPC and Intel platforms.
ASTZOOMMAP in ATOOLS creates a new AST ZoomMap.
- ASTDISTANCE in ATOOLS calculates the distance between two points, given an AST Frame.
- PROVMOD in KAPPA allows provenance information in an NDF to be modified.
- PROVREM in KAPPA removes selected provenance records from an NDF.
- REGIONMASK in KAPPA masks out regions of an NDF, similar to ARDMASK, but uses AST Regions instead of ARD descriptions.
- WCSSLIDE in KAPPA applies a translational correction to the World Co-ordinate System (WCS) of an NDF and may be used to correct pointing errors.
- UNMAKECUBE in SMURF creates simulated time-series data from a given sky cube, using existing time-series cubes as templates.
Changes to Applications
- New command ASTDISTANCE finds the distance between two points in a given Frame, along the geodesic curve that joins the two points.
- A memory handling bug in FINDOFF has been fixed. This caused the ORAC-DR pipeline to crash.
- NDF2FITS supports the propagation of provenance information to FITS headers. There is a choice of generic propagation that attempts to propagate all the information, or to write CADC-specific headers, or to exclude provenance.
- NDF2FITS now handles extensions containing only NDFs by adding a dummy FITS sub-file that retains the name and type of the wrapper structure.
- FITS2NDF processes SMURF-package data better, permitting a roundtrip via FITS, perserving the original data structures, save for some additional FITS headers.
- The creation of AXIS structures by SPECX2NDF can now be suppressed using its new AXIS parameter.
- NDF2FITS is now much quicker when writing out WCS information that includes a large table of values.
- An option has been added to FINDCLUMPS to control whether a background is subtracted before calculating clump sizes. Switching this option off allows more accurate comparison with the IDL version of CLUMPFIND, but causes reported clump sizes to depend on the background level.
- A problem using sort function when compiled using g95 has been fixed (this caused DAOPHOT to abort).
- New script trendview plots multiple spectra from a cube overlaying fitted trends and spectral-feature mask. It enables assessment of baseline subtraction and feature masking.
- peakmap and velmap now have options to inspect and approve the fit at each spatial pixel, and to log the Gaussian fit parameters and their errors in a Small Text List file.
- DSBSPECTRUM is now recognised as a valid spectral domain in velmap.
- peakmap: a bug has been fixed that caused the peak-height errors to be stored within the output NDF 's VARIANCE component, instead of the variances.
- velmap: the velocity variance (instead of the error) is stored in the output NDF's VARIANCE; the WCS Frame is reset before re-plotting in final individual-pixel refit loop to prevent the wrong spectral co-ordinate being plotted; and the logic for deciding whether to create or switch WCS Frames is disentangled.
- A new command-line and startup option -pixel_indices is available. This changes the X and Y continuous readout displays to show integer NDF pixel indices (include the origin), not grid co-ordinates.
- Contour and grid overlay lines can now be drawn in dotted and dashed styles.
- All dialogues for opening images, local catalogues and cubes have been enhanced to offer a new "Browse" facility. This allows you to view any FITS extensions (HDUs) or NDFs also present at the same level in the container.
- Local catalogues can now also have the HDU number given as part of their name.
- Cubes which at the same level as the currently accessed cube can now be selected using a chooser like that already available for images.
- When opening an HDS container file that does not have an NDF at the top-level a search for cubes at the next level is now made and the first NDF found is opened (this has always been the behaviour when opening images).
- New hot keys "1", "2" and "3" have been added to the 3D display for pointing an axis out of the screen.
- The cube toolbox can now display the data variances as standard deviations.
- A new command-line and startup option -font_scale has been introduced. This sets the scale of non-pixel based fonts or adjusts the padding around pixel-based ones.
- Choosing fonts is now easier using a font chooser dialogue available in the startup options.
- When saving a FITS image with a new world co-ordinate system the old behaviour was to always write the astrometry in the format of the image. This has been changed so that FITS-PC astrometries will now be written in the standards complying FITS-WCS format (see SUN/211 for the details of this terminology).
- A snapshot feature that captures the main window display to a graphic file has been added. Supported formats are GIF, JPEG, PNG and TIFF.
- A new interaction for key "c" has been added to the 3D display. This centres the scene without changing the zoom.
- Local time "LT" has been added as a possible timescale for extracted time-series spectra.
- A description of the type of each cube axis will now be shown in the cube toolbox.
- When sending spectra to SPLAT it should now be possible to restart GAIA without restarting SPLAT (previously spectra open in SPLAT would not be superseded by new spectra sent from GAIA).
- Contour levels can now include missing values (previously attempting this caused a Tcl error).
- The AXIS component of an NDF cube will now be propagated with any extracted spectra. Previously this didn't happen so saved spectra could lose their AXES co-ordinates.
- The data limits used for spectral extraction will now be updated when either a slave's master starts an extraction, or the 3D scene starts moving the spectral extraction shape (region or line).
- A bug displaying the extraction limits on the spectral plot which caused the limits to be reversed when moving the graphics, has been fixed.
- When creating a channel map from a cube some of the combination estimators require more than one channel to operate (median for instance), this problem is now reported avoiding an uninformative error message.
- When capturing an animation to a GIF the main window must be clear of all overlapping windows. This requirement is now reported as a warning rather than a generic "too many colors" error message.
- A bug handling integer images has been fixed. Previously contouring, XY extraction and region statistics were broken for this data type (since Version 4.0).
- The blink toolbox should now handle the case when RA and Dec are not axes aligned more effectively.
- A problem reading very large FITS extensions in NDFs has been fixed. This should speed up startup times.
- The colour ramp shown at the bottom of the main window will now correctly update when a new colour or intensity map is selected.
- A bug in the 3D toolboxes that meant reloading a cube with the same name under certain circumstances would fail. This issue should now be resolved (although re-using files with the same name remains unsupported).
- A problem reading co-ordinates from FITS tables has been fixed. These were being read with insufficient precision leading to alignment error when displayed.
- A bug that stopped cube rebinning using a factor of 5 has been fixed.
- New command PROVREM removes selected provenance records from an NDF.
- New command PROVMOD allows provenance information in an NDF to be modified.
- New command REGIONMASK allows for masking out regions of an NDF using a textual description of an AST Region.
- New command WCSSLIDE applies a translational correction to the World Co-ordinate System of an NDF and may be used to correct pointing errors.
- COLLAPSE has a COMP parameter to permit the variance to be collapsed.
- FFCLEAN was modified to operate on cubes, either by processing the cubes as a set of independent one-dimensional spectra, or as a set of independent two-dimensional images. It can now store the calculated noise level in the output variance array.
- FITSMOD has a new editing option "Amend". This will write a new header if the selected keyword does not exist, but will only modify an existing header. Thus it is now not necessary to know whether or not a header exists.
- MAKESNR made faster.
- MFITTREND has a cubic-spline fit option. There is a choice of fit: interpolation through specified or equally spaced knots, or smoothing at automatically generated knots.
- MSTATS now offers a wider selection of statistics to combine the supplied NDFs. There is optional variance weighting. It now uses NDF blocking to avoid excessive memory requirements.
- PROVSHOW has a new parameter SHOW that can be used to select the ancestors that are to be displayed (root ancestors, direct parents, or all ancestors).
- SETBB has been changed so that the bit mask can be specified by a list of quality names.
- Quality names created using SETQUAL can now be flagged as "read-only", meaning they cannot be removed using REMQUAL.
- WCSMOSAIC Parameter GENVAR now allows a null value, which requests that a value of FALSE be adopted if and only if all the input NDFs have variance components.
- Parameters that require one or more WCS axes to be specified now additionally accept generic names of those axes present. These generic names are: "SPEC" to select the spectral axis, "TIME" to pick the time axis, and "SKYLON" and "SKYLAT" to choose the sky longitude and latitude axes respectively.
- Truncate trailing blanks from created character components.
- REGRID no longer flips the RA axis in an (RA,Dec) image.
- WCSMOSAIC now sets the output pixel origin correctly.
- FITSTEXT can now handle lines that have white space that extends beyond column 80.
- The available observatories list has been updated to match those supported by the SLA library.
Improvements in ACSIS data reduction, including iterative coadded cube production, quality-assurance tests, and improved baselining. Off-line data reduction will be greatly improved over the standard "summit" pipeline by adding "<tt>-recsuffix ADV</tt>" to the command-line. This uses the advanced recipes and produces better data products.
- ORAC-DR Tk window can now be usefully resized.
- New ORAC::Group::members_inout() method that replaces ORAC::Group::inout(). ORAC::Group::inout() has been modified to do for group file names what ORAC::Frame::inout() does for frame file names.
- New orac_say() function that does what orac_print() does, but appends a carriage return to the output string.
- New get_prim_arg() function to reduce lines-of-code needed to retrieve primitive arguments.
Various bugs popped up when switching from internal header translation to external header translation using Astro::FITS::HdrTrans. Most of these bugs have been fixed.
- A bug that stopped the PHOTON parameter from being greater than 2 (excluding the options for estimating errors from the data variance and Gaussian estimation).
- Various coding errors that could result in the failure to output information and error messages have been fixed (mainly in the optimal-photometry code support).
- The FWHM value is now reported correctly for the PSF star in optimal photometry (previously underestimated by a factor 1/1.665).
- POLPACK applications now propagate provenance information.
- New command UNMAKECUBE added, which creates simulated time-series data from a given sky cube, using existing time-series data as templates.
- The TIMESORT command can now sort data across multiple sub-scans, and can also remove dead detectors.
- Refined support for polarisation data in MAKECUBE (analyser angle for each bin is now displayed, WCS handling improved, all output polarisation data now placed ina single container file).
- MAKECUBE now uses the size of the Airy disk to determine whether spectra are spatially co-incident.
- MAKECUBE now uses the MAP_PA FITS header to determine a default value for the CROTA parameter, rather than using a fixed default value of zero.
- MAKECUBE has a new parameter called OUTFILES which is the name of a text file into which MAKECUBE writes the names of the output NDFs.
- MAKECUBE (in verbose mode) now gives a continuously updated display of "percentage done" when processing large files.
- TIMESORT can now merge data from two or more sub-scans.
- TIMESORT now has an option to fill the variance arrays in the output NDFs with values derived from the Tsys values in the input NDFs.
- TIMESORT now has an option to exclude or include data from a specified set of detectors.
- TIMESORT now has an option to purge dead detectors from the output NDFs.
- TIMESORT has a new parameter called OUTFILES which is the name of a text file into which TIMESORT writes the names of the output NDFs.
MAKECUBE can now be asked to share out the work between different cores on a multi-core processor, thus reducing the overall run-time. MAKEMAP also benefits from these options if run with METHOD=REBIN. Multi-threaded mode is experimental (please email email@example.com with any problems or concerns) and is enabled by setting the SMURF_THREADS environment variable to the number of threads you wish to use. This mode is turned off by default.
- A new command-line option --debuglevel to control the messages output on the terminal has been introduced. This aids in diagnosing problems.
- The errors of a spectrum can now be plotted as the spectral line, not just as error bars.
- The line profile fits now include error estimates for the fitting parameters.
- The SSAP query options have been extended to include TIME, FLUXCALIB and WAVECALIB.
- The SSAP query options now allow the object name to be passed as part of the object, without looking up an RA and Dec. This is suitable for searching for Solar System objects.
- The query SSAP server dialogue now offers "select all" and "remove unselected" options.
- An error in the derivates used when fitting the Voigt profile has been corrected. This should give closer fits in some cases.
- MIME types from SSAP queries may now include trailing attributes.
- Saving the result spectrum of binary maths operations to FITS format should now work. Previously an invalid FITS file was created.
Changes to Libraries
It is strongly recommended that all applications that use any libraries be rebuilt from source.
Many changes have been made to the ast.h header file. All source code that includes <tt>ast.h</tt> must be recompiled.
- AST is now thread-safe.
The TimeMap class can now represent Local Time.
- The Mapping class has a new attribute "isLinear".
- Added regular expression facilities for string splitting, matching, substitution and parsing.
- Add option to astResample that allows output pixels to be left unchanged if an interpolated value cannot be obtained for them.
CmpFrame class changed so that by default, a CmpFrame will match any other Frame that contains a subset of the axes in the CmpFrame. For instance, a (detector index,mjd) CmpFrame will now match a ((velocity,detector index),mjd) CmpFrame.
- The final field of a formatted sky co-ordinate may now include an exponent.
- Adjust the spatial CRVAL and CRPIX values written by the FITS-CLASS encoding to refer to the centre of the first pixel if the spatial axes are degenerate.
- Modify fitschan to produce correct absolute co-ordinates when writing an offset co-ordinate system to foreign FITS.
- Correct the use of VELO-LSR when reading a FITS-CLASS header.
Improve implementation of astMapSplit provided by the CmpMap class so that it manages to split more Mappings succesfully.
FitsChan is now much quicker when reading or writing very large FITS headers.
New routine atlWcspx that creates an AST FrameSet describing a SPECX dataset.
- Support for using EMS together with POSIX threads has been added. SSN/4 has been updated to describe this change. This is enabled by default on platforms with POSIX threads support.
- A new routine EMS_GTUNE has been added, with a C version emsGtune.
- The emsTune routine has been deprecated and replaced by a new routine emsStune. This also returns the current value of the tuning parameter.
- A memory overrun bug storing and retrieving parameter names has been fixed. This fixes a problem in GAIA.
- A bug when concatenating message tokens inside a mark has been fixed.
- An issue in HDS that stopped the loading of shareable libraries by Java under OS X 10.5 has been resolved.
- Extend IRQ facilities to allow more flexibility in usage of quality names.
Most of MERS has been rewritten in C from the original Fortran. This is part of an ongoing effort to make MERS thread-safe.
- A bug that caused a graphics X window created by ORAC-DR to have no size has been fixed.
The STARJAVA applications and libraries were built using Java version 1.4.2. A version of the Java runtime environment of at least this level is required to make use of them.