nip2 displays wrong values for uint image, instead displaying values interpreted as int

[erdmann]

Hi John,

I have a vips format (.v) single-band file that's full of '0xFFFFFFFF' values, i.e., 2**32 - 1 = 4294967295. The band format is 'uint', and both libvips and nip2 correctly report this. Values reported in cells, such as max(A1), are also correct.

The bug is in the image viewer, which is apparently incorrectly interpreting these values as 32-bit signed 'int' values, i.e., as -1 since (int)0xFFFFFFFF == -1. This problem appears both in the numerical display of the pixel values under the mouse cursor as well as in the display of the brightness since all such pixels appear as black. In other words, it's apparently not just a problem with "%u" instead of "%d" in a sprintf statement somewhere. The thumbnail is also incorrectly showing the image as black instead of white.

Many thanks in advance.

[jcupitt]

Yes, I see this too. I'll have a look. Thanks for the report!

[jcupitt]

There was a libvips bug too, ouch. Fixed in git master, thanks!