John Cupitt

Possible sample image: https://gitlab.gnome.org/GNOME/eog/uploads/e4bf3ae5d2b45afb9590250b4df3cacb/rgba16f.tif

I had a quick look and because of the structure of tiff2vips.c this is harder than it sounds. We need to add an extra "unpack to native" layer under all...

Hi @xamir82, You're right, this is a libvips bug. I guess it should look for `,([A-Za-z]|])` as the argument separator when parsing those strings. As a workaround, you can call...

Hello @jstkdng, That's very cool! Did you see https://github.com/jcupitt/vipsdisp ? It's not doing the same thing, but maybe there are some ideas you can use. Try it on a large...

> well, the work of loading the image into libvips and then creating an xcb_image has to be done at some point, either after the event or before the event....

> I still need to make some pre-processing on the image, wouldn't using sink_screen load the image again? What kind of preprocessing do you need? It should all be inline,...

Using the truncated jpg in the linked discussion, before this patch I see: ``` $ vips copy trunc4.jpg x.png (vips:283787): VIPS-WARNING **: 11:58:21.438: error in tile 0 x 856 VipsJpeg:...

Huh this is a strange one, you're right. I made a branch with this fail-on change, plus your remove seq from reduce patch: https://github.com/libvips/libvips/tree/remove-seq-from-reduce And I can't provoke an out...

Oh haha this fails instantly: ``` $ vips black x.png 10000 10000 $ VIPS_STALL=1 vips reducev x.png x.v 2 vips_worker_work_unit: stall done, releasing y = 0 ... (vips:457884): VIPS-WARNING **:...

I tried with a real image (not just black): ``` $ vips crop st-francis.jpg x.jpg 0 0 10000 10000 ``` And verified that this produced the correct result: ``` VIPS_STALL=1...