Mateusz Łoskot

Thanks for the report. Sorry, but I only skimmed it. > `printf("TONY: BEG f.min_v_: %ul f.max_v_: %ul \n", f.min_v_, f.max_v_ );` The channels can be signed too, but I believe...

> `TONY: BEG f.min_v_: 4294967168 f.max_v_: 127` For the `f.min_v_` above, which is 8-bit value of `-128`, something is casting it to unsigned 32-bit integer: `2^32 - 128`, what becomes...

@trex58 Yes, it looks like a bug somewhere in packed and/or bit-aligned images, possibly big-endian related. I am fluent yet about internals of those types, so it is hard to...

It is a good idea.

@stefanseefeld Since, - is not assigned to [The Grand Merge](https://github.com/boostorg/gil/milestone/1) - no work has been done for this so far, AFAIK I'm removing it from Boost 1.68 and postponing. Feel...

While cleaning the headers lately, I've added some `TODO` comments about dependency on the toolbox and dynamic_image extensions.

@marco-langer Boost.Geometry's version allows pairs of integers ``` using B = boost::geometry::select_most_precise::type; ``` while your variant requires one of the two must be float-point. Is that intentional? @sdebionne >Shall we...

> This non-commutativity is either a.) just unintuitive or b.) a potential source of bugs. A very good point. > While implementing my solution I too came to the same...

@marco-langer How about @sdebionne 's #204 ? n.b. it's also a follow-up to #363 discussion, so could be fixed/closed by this PR too.

FYI, I reported the issue here, https://developercommunity.visualstudio.com/t/INTERNAL-COMPILER-ERROR-compiling-std::v/10011526 > Which make me wonder why use /std:c++latest? Isn't this asking for troubles 😀 ? Yes, `/std:c++latest` needs to be disabled. This is...