From: Craig D. <zo...@us...> - 2011-11-20 06:32:58
|
This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "pdl". The branch, master has been updated via 1fce6d43ac87bca690e8581c6291a55af235020f (commit) via 6dfce88bd66b88381198381413296d54fb12023d (commit) from d5f447263f778abf5dfd16e9dd4ad27d951b85fb (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit 1fce6d43ac87bca690e8581c6291a55af235020f Author: Craig DeForest <zowie@Clio-2.local> Date: Sat Nov 19 23:30:04 2011 -0700 Fixes bug 3299611 - obscure fits I/O bug. Replaced a simple "clump(-1)" operation with a slightly-more-expensive "->copy->reshape($p->nelem)" operation, which forces regularization in memory. Since disk I/O is (thought to be) slower than memory operations, and since clump often copies memory anyway, this seems a small price to pay. The alternative would be to dive into XS and access the dimincs fields of the subject pdl directly, but it is not clear there would be a significant performance advantage in the usual case. commit 6dfce88bd66b88381198381413296d54fb12023d Author: Craig DeForest <zowie@Clio-2.local> Date: Sat Nov 19 23:12:22 2011 -0700 Fixes bug 3428356: PDL::Transform::map output FITS header is slightly off The problem was an operation order: a half-pixel coordinate offset (used to find the pixel coordinate of the corner of the outermost image pixels) was happening *before* the subsample scaling operation (a multiplication) rather than *after* it. Fixed. Thanks to Derek for finding and reporting this. ----------------------------------------------------------------------- Summary of changes: IO/FITS/FITS.pm | 2 +- Lib/Transform/transform.pd | 4 ++-- 2 files changed, 3 insertions(+), 3 deletions(-) hooks/post-receive -- pdl |