Mercurial > octave-nkf
view .hgignore @ 20373:a7770c66cb3d
Saturated red channel on RGB image, makes imread() return logical (bug #41584)
* libinterp/dldfcn/__magick_read__.cc (get_depth): this function tries to
second guess GM when reporting the actual bitdepth of an image in the file
(see comments for full discussion). We assumed we could check only red
from RGB but if the channel was saturated, it would return depth of 1, so
we really need to check all 3 channels. Unfortunately, 3 saturated channels
may still lead to images incorrectly read as logical.
* scripts/image/imread.m: add test for this bug.
author | Carnë Draug <carandraug@octave.org> |
---|---|
date | Fri, 10 Jul 2015 16:46:24 +0100 |
parents | dc76e3909d36 |
children | c16c446204e5 |
line wrap: on
line source
syntax: regexp # The recurrent (^|/) idiom in the regexps below should be understood # to mean "at any directory" while the ^ idiom means "from the # project's top-level directory". # gnulib makes these silly backup files .~$ # This directory gets created by gnulib during the build. # It contains no Octave files. ^libgnu/* # This directory mostly contains cruft during build time, but it does # contain some Octave code, so we gotta be a bit more careful about # what we ignore here ^m4/(?!ax_).+\.m4$ # Emacs tools create these (^|/)TAGS$ (^|/)semantic.cache$ (/|^)Makefile\.in$ ^INSTALL$ ^aclocal\.m4$ ^build-.*($|/) ^configure$ ^autom4te\.cache($|/) ^config\.in\.h$ # e.g. doc/faq/OctaveFAQ.info # doc/interpreter/octave.info-4 ^doc/.*\.info(-\d)?$ ^doc/\w*/stamp-vti$ ^doc/\w*/version\.texi$ ^doc/interpreter/images\.mk$ # e.g. liboctave/operators/smx-op-inc.mk ^liboctave/operators/\w+-op-\w+\.mk$ ^libinterp/dldfcn/module\.mk$