Discussion: SFML 3.0

[NQNStudios]

I just saw in the SFML Discord that they put out the first release candidate for SFML 3:

This response is great:

😆

---

So we've talked about things we dislike in SFML 2, and porting to SDL instead has been mentioned. Personally I think changing backends should be considered a monstrously serious investment that we can't afford to make with just the current number of contributors to fix the inevitable influx of new bugs that would come from that. Being pretty locked into SFML, cool new features would be nice--I'm looking forward to reading the changelog and seeing if there's anything new in it that would benefit cboe. And maybe I'll try building against this release candidate in a new branch, if I see anything worthwhile.

[NQNStudios]

In the changelog:

New and improved event handling APIs ⌨️

[NQNStudios]

It requires C++17.

[CelticMinstrel]

I'm definitely very reluctant to bump the C++ standard requirement. I'm pretty sure C++17 would mean I can't build in Xcode 4 anymore. (Though, there's a good chance the current code already doesn't build in Xcode 4, since I haven't been checking it recently.)

I think it might boil down to "we should support building against either SFML2 with C++11 or SFML3 with C++17", but I don't know how much work that would actually be. I'd at least need to have my old Mac set up before I can really think about it.

[NQNStudios]

Apparently they're going to backport the bug fixes to SFML 2.6.2 anyway so maybe we'll get all those without doing the bump.

[CelticMinstrel]

Sounds like we have plenty of time to decide.