[account42]

> In 2016 I wanted to change the type universally to just typedef struct Curl_easy CURL; … as I thought we could do that without breaking neither API nor ABI.

This seems to be the obvious solution and how most libraries define their opaque handles. It doesn't break a guaranteed API and it doesn't break the ABI any more than than only using it when building the library - and you can check that it doesn't break the ABI on any platform where you want to guarantee ABI stability.

[mort96]

In the real world, we often want to avoid breaking people's code even when people rely on something that's not guaranteed by the API docs. It seems like Daniel's goal isn't to only be API-compatible in a technical sense (namely that perfectly written code which carefully avoids using anything in a way that's not explicitly guaranteed to work), but rather to avoid breaking people's existing code. I can respect that.

[account42]

Yes but that is always a balancing act if you want to make any change at all since users can theoretically depend on any possible implementation detail or even on outright bugs.

[mort96]

Correct. It's always a balancing act. That means hard rules like "we will do any change which doesn't technically break any promises explicitly made in the documentation in a patch release" aren't appropriate, it's always a value judgement (and sometimes you get it wrong and revert the change once you realize that people were affected more than expected, as happened in the case of curl).