On 4/2/24 18:24, Peter Dimov via Boost wrote:
Andrey Semashev wrote:
So my main point is that "old code uses old Boost" is a myth in practice.
It's not a myth; old code does use old Boost. But at the same time, old code often needs to upgrade its Boost if it upgrades something else, such as compiler or OS, because the old Boost sometimes breaks.
If the old code can be rebuilt with the newer Boost, that's not what I call "old code uses old Boost". It's when you can't rebuild, or put another way, when there is a non-trivial amount of work that needs to be done to port the old code to the newer Boost. And the myth is that such "old code uses old Boost" state is practically sustainable for that old code. Either the old code gets phased out (which can be painful for its users) or it has to be ported to the newer Boost (or away from Boost) to remain relevant and available to users.