TFIT Release Notes Version 1.2, 10 Mar 2008: r394 This release restores the ability to use multiple kernels in first pass processing, and eliminates the printing of informational messages and unimportant warnings during the convolve stage. Changes include: - Cleaned up some memory leaks & modified string handling in the fitter stage; also disabled the ability to run the fitter with command-line arguments. - Turned off PyFITS warnings during the convolution stage (they're turned back on afterwards); messages are no longer printed for acceptable anomalies. - Added a posframe parameter to the parameter file that specifies whether the lookup file contains pixel coordinates in the hires frame or the lores frame. If the former, multikern uses the cutoutcat; if the latter, it uses the lrfcat. Multikern now creates the output template directory if it doesn't already exist, and does not need a flag file in order to run. - Made tfit_diags.collage more robust against missing template files or templates that for some reason don't fit properly into the collage: a message is printed & the task continues, instead of aborting. - Corrected some errors in the utility scripts section of the user's guide; updated user's guide for v1.2. Version 1.1, 24 Jan 2008: r353 This release provides the ability to save templates (cutouts convolved with the kernel) at the pixel resolution of the HR image. It also includes a utility script that can use these templates and the cutouts to construct scaled collages useful for PSF investigations. Changes include: - added parameters to tfit_pipeline.param to specify saving the templates at high resolution - a new keyword TBINFACT is now written to the template header, specifying the binning factor used when the template was created. - made pipeline more robust when optional parameters are missing from the .param file - added utility script psf_collage.py to build matching collages from the cutouts and the hr templates - installed the utility script make_collage.py, that can make a collage without re-running the entire diagnostics stage. - various internal improvements (transparent to user) in path and catalog handling Known issues include: - Use of multiple kernels in first pass processing is presently broken (ticket #11). This will be fixed in the next release as part of the planned enhancements in ticket #80. - Not all path handling is platform independent (ticket #78). This ticket was closed without being fixed: problems only arise on Windows, and we do not plan to provide support for this package on Windows. Version 1.0, 21 Dec 2007: r322 This release corresponds closely to the version of the software described in Laidler et al. 2007, PASP 119:1325. Changes include: - the numpy array package is now used in place of the deprecated numarray package - a new idmap diagnostic product is now produced by the diags stage - additional utility scripts are provided - various internal improvements (transparent to user) in path and catalog handling Known issues include: - Use of multiple kernels in first pass processing is presently broken (ticket #11). - Not all path handling is platform independent (ticket #78) TFIT is managed as a component of the AstroLib project: http://astropy.scipy.org/astropy/astrolib/wiki