Mateusz Łoskot

@simmplecoder Thanks for the very detailed tests. I don't have any immediate conclusions to share. I'd however suggest to try the current develop branch, where Boost.MPL has been replaced with...

@sdebionne FYI, we've started discussing the benchmarks https://lists.boost.org/boost-gil/2019/05/index.php Your insights are very welcomed!

Here is another micro-benchmarking library that some of us may like to consider and try out https://github.com/martinus/nanobench

@psbro I think the task in this issue is self-explanatory (with help from reading the linked references, GIL I/O docs and source code and other online matrials about the format...

I'm also suspicious about the `crc_32_type` leaving slight chance for misbehaving `img_view` (or compiler tripping over it) here ``` boost::crc_32_type checksum_acumulator; checksum_acumulator.process_bytes(img_view.row_begin(0),img_view.size()*3); ``` I'll try to reduce the case during...

I've changed the `test/image.cpp` to **temporarily** skip the four failing test case.

### Test 1  `image_crc_minimal` runs series of test cases with `NxN` images where `N = {2, 3, 4, ..., N+1}` each time performs 3 steps: - create empty `NxN`...

@stefanseefeld > What are the hex numbers between the lines starting with "Checksum" ? This is hex-encoded binary of test image > What does the last screenshot above actually show...

@stefanseefeld > An interesting (though not necessarily important) question is whether this bug also exists on the master branch, as it doesn't seem to be related to the IO extension....

> Unless you think you could fix the issue quickly No idea really, it's still a cowboy shoot-out. > I suggest you add that test to GIL's test/ (or perhaps...