Re: UUID design discussion

From: Vinnie Falco (vinnie.falco_at_[hidden])
Date: 2024-04-25 16:07:27

On Thu, Apr 25, 2024 at 9:01 AM Andrey Semashev via Boost <
boost_at_[hidden]> wrote:

> I think, the only input character type we should allow
> is char. And we should not care what encoding it is or whether it is
> valid or not. IOW, we should take it as an opaque sequence of bytes.
>

I don't know much about the specifics of Boost.UUID but this solution
appeals to me because it pushes all responsibility for figuring out Unicode
or encoding equivalences onto the caller where it belongs. Separation of
concerns is better than forcing every library that interacts with strings
to have to solve these same problems over and over again (usually
inconsistently).

Thanks