Version: (using Devel) Compiler: gcc 4.3.0 OS: Linux Installed from: Compiled sources I'm viewing this large PNG image: http://hemswell.lincoln.ac.uk/~padams/kde-413.png When I set a zoom level >= 84%, nothing gets rendered and I can only see a blank white document. At lower zoom level, after a few CPU-intensive seconds, the document renders correctly. Even at 83.5%. Some libpng version information: gaston@kiwi:~$ ldd prefix/bin/okular | grep png libpng12.so.0 => /usr/lib/libpng12.so.0 (0xb6622000) gaston@kiwi:~$ ls -l /usr/lib/libpng12.so lrwxrwxrwx 1 root root 18 2008-03-25 20:49 /usr/lib/libpng12.so -> libpng12.so.0.15.0
This is the same case as bug #173681, even if does not look at a first glance. The reason behind both is the same: Okular does not render (by choice) "big pages" to avoid to take too much memory. *** This bug has been marked as a duplicate of bug 173681 ***
(Also, you might want to use a real image viewer, as Okular is not an image viewer.)
Heh. Either you support a feature or you don't. If you want to support PNG's but only small enough or at small enough zoom, then you need to do something else than just failing silently. Maybe an error message in the GUI explaining the situation and proposing instead to open the image in the default application for this mimetype (which I suppose would be Gwenview).
okular launching gwenview? why would you want to do that, if you use okular you get what okular provides, if you don't like it, don't use it and use gwenview :-)
ok ok that's your call :) but in any case: if okular claims to be able to open image files, and it turns out that it can't open all of them at any zoom level, then user.isConfused(). So at least, a user-visible error message, or something like that. I then thought that, seeing this message, the next thing that the user would do, would be to open the same file with the default image viewer. So I thought, why not just allow him to do it in 1 click from the error msg. But that's just my 2 cents :)
Since Gwenview can render those images, would it be possible to share the code wich makes this possible (as noted in bug 173681, this also hits PDF files).