It looks like the tag type was previously used only for type traits. If so, it's necessary to rewrite type-traits and remove that type.

Types length and all the related logic are counted in bytes (8-bit) for now. For the purposes of versatility we better switch to bit-oriented counting.

Marshalling containers currently seem to be very similar to the boost::container ones. It would be great to completely substitute them.

• array_view
• static_vector
• static_string
• string_view
• static_queue (currently not working, not used)

There is a great idea how to simplify field options:
422ddf4e2edd55b514533f81855b3b32c227b16e

Currently, marshalling module contains a lot of things, which ways of usage are very limited to some network tasks. It will be better to move all this functionality to a separate module.

Currently, we only have big-endian and little-endian.
little_unit_little_bit and big_unit_little_bit are required to be added to fulfill all the current needs.

As I see it now, requires extending the processing namespace with corresponding functions.

There is a lot of outdated links and mentions in the doxygen documentation right now. It should be fixed.

At the moment, pack algorithm is part of long-running backlog. It is placed in the crypto3::detail namespace. We must substitute it by the marshalling pack. To be sure, that we made correct substitution, we need to move pack tests to marshalling and ensure, that there is no probplems with that.

Module has a lot of internal type traits. They all have to be reduced to the same form.

We want to implement CI/CD pipeline, where different submodules of crypto3 will be tested with respect to each other PRs, if all those PRs are linked to the same issue using [SyncWith org/repo#issue] syntax in PR's title.
Current issue will be used for testing of this functionality