> On Sept. 10, 2012, 10:48 p.m., Allan Sandfeld Jensen wrote: > > kimgio/webp.cpp, lines 67-73 > > <http://git.reviewboard.kde.org/r/106300/diff/2/?file=82722#file82722line67> > > > > This seems needlessly explicit, why not just copy the whole block into > > QImage?
QImage::scanLine docu says "the pixel format depends on the byte order on the underlying platform." So to be sure to create correct rgba tuples, I explicitely pass the values to qRgba/qRgb > On Sept. 10, 2012, 10:48 p.m., Allan Sandfeld Jensen wrote: > > kimgio/webp.cpp, lines 92-100 > > <http://git.reviewboard.kde.org/r/106300/diff/2/?file=82722#file82722line92> > > > > Same as for decoding. Why copy it color by color like this, when the > > formats are identical? Same answer as above: scanLine data depends on the machines byte order, but WebP has always the same data representation > On Sept. 10, 2012, 10:48 p.m., Allan Sandfeld Jensen wrote: > > kimgio/webp.cpp, line 127 > > <http://git.reviewboard.kde.org/r/106300/diff/2/?file=82722#file82722line127> > > > > Not really required, but it would be nice to support the Size option. > > It is quite often used, and should be available through WebPGetInfo() ok, added > On Sept. 10, 2012, 10:48 p.m., Allan Sandfeld Jensen wrote: > > kimgio/webp.cpp, line 159 > > <http://git.reviewboard.kde.org/r/106300/diff/2/?file=82722#file82722line159> > > > > Should it also require "VP8 " in byte 12 to 16, to protect against new > > unsupported versions? > > > > For some reason I have code that checks the file is atleast 32 bytes > > long, I think that is because it is what WePGetInfo requires. not really. The spec says that the WebP header is 12 bytes and contains what I check. There is an Extended file format which starts with a ‘VP8X’ chunk (can currently contain "VP8 " or "VP8L") but as I do not interpret that VP8 data, why restrict the reader to it? Maybe a future libwebp supports something else than VP8 after the WebP Header ... - Martin ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: http://git.reviewboard.kde.org/r/106300/#review18829 ----------------------------------------------------------- On Sept. 1, 2012, 9:23 p.m., Martin Koller wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > http://git.reviewboard.kde.org/r/106300/ > ----------------------------------------------------------- > > (Updated Sept. 1, 2012, 9:23 p.m.) > > > Review request for kdelibs. > > > Description > ------- > > This patch adds support for the new Google invented WebP image format. See > https://developers.google.com/speed/webp/?hl=ru > > The patch is missing a cmake rule to make it optional, though, but I sent a > mail to KDE-core list in the hope someone knows... > > > This addresses bug 267365. > http://bugs.kde.org/show_bug.cgi?id=267365 > > > Diffs > ----- > > kimgio/CMakeLists.txt 26329c0 > kimgio/webp.cpp PRE-CREATION > kimgio/webp.desktop PRE-CREATION > kimgio/webp.h PRE-CREATION > mimetypes/kde.xml a82b87c > > Diff: http://git.reviewboard.kde.org/r/106300/diff/ > > > Testing > ------- > > some KDE apps, including read/write with a modified kolourpaint (to be able > to change the quality) > > > Thanks, > > Martin Koller > >