It's worth being concerned about losing the lessons learned from Qt, but this is a kind of false dichotomy between the old and the new.
With that said, I don't think those lessons will be lost anyway. Qt can't (and shouldn't!) last forever or remain in its current form. C++, though it will live for a long time yet, will eventually fall out of favor, and large sections will eventually be purged from Qt as the project matures. The man-years that went into the portions that were, or some day will be, removed have taught Qt developers and users a great deal.
Qt developers and users will bring knowledge and wisdom to new projects that will compete with Qt (directly or indirectly). Some projects will survive and some won't. Some will be much the same as Qt, while some will be completely different from the start. Conrod, as it happens, is an approach to a different kind of GUI development (see this comment: https://news.ycombinator.com/item?id=8244224).
This, of course, doesn't include developers who do their own things without ever using Qt. Hopefully, they will encounter blog posts and articles about some of those lessons, as well as attracting developers who've learned some of these lessons.
This isn't a case of getting away with reinventing the industry. In this nascent stage, computer science, with all the renegade hackers and developers, is still a necessary occurrence when new languages and frameworks are invented.
C++11 and C++14 have done a lot to enable users to write clean, efficient code. I probably would not be using C++ today if I were forced to write in C++03 style. I am still not convinced that C++ has gotten any easier to learn over time. But for those who know how to use it well, no language ranks better in allowing users to write clean, efficient, and portable code.
No, what worries me much more is the fact that there is no antialiasing on the "Envelope" in demo video.
Besides, probably it will surprise or even scare you, but Qt isn't the only, the first or the last GUI framework, there're many more, and new are created all the time. …And not even single actually good one invented still, but that's another topic. So there isn't any reason to think of that Conrod thing as something more than "just another one GUI framework". At least, until it would turn out to be somehow superior to ones that already exist. And if it will turn out to be superior to, say, Qt, then Qt should go to trash and I couldn't care less about all that "man-centuries" of yours. But, realistically, it's unlikely to happen, so everything we could hope for is some less complicated, useful for just some subset of GUIs micro-framework. In fact, it would be already very nice by itself, since it would spare Windows-developers of the pleasure of 30Mb cross-platform "Hello worlds" using Qt.
Qt is excellent for its epoch, but if you've read the source you will know that we should not treat it as sacrosanct. Algebraic types and explicit lifetime declarations and macro syntax extensions could all do wonders for it.
I imagine that as soon as Rust hits something like version 1.5, we will start to see semi-automatic converters from C/C++ to Rust, something similar to the C-to-Go effort underway to get to a self-hosted Go compiler.
Qt will probably take 5 years to be ported in a decent way, but I am pretty sure that a sizeable part of that codebase will be translated without too many problems, given its relative clarity.
Does anyone consider Ada another contender to C++? Statically strongly typed, safe manual memory management, wide-spectrum, hard real-time & embedded capable.
If Rust wants to become a real systems programing language, it needs OS vendor support, not HN cheering.
I have worked for quite a few real-time safety-critical systems, including one of the domains you mentioned. I have ex-colleagues who now work for other such domains. I assure you, that Ada is rarely ever seen. It's just C all the way, and usually running without an OS. I used to see job openings for Ada programmers some 15-20 years ago, but I haven't seen anyone use it in any of the firms I worked at. The only thing that seems to be changing these days is a creeping in of Matlab/Simulink-generated C code.
On one hand, I'm enthused that Rust, arguably the only serious contender to C++'s throne, is gaining traction.
On the other hand, I'm scared that respectable projects like Qt (~6M lines of C++, man-millenia of work!) will be considered "obsolete" by the coming generation and another cyle of wheel reinvention will begin, tossing away man-centuries worth of polished, working code on the sole ground that it's C++.
It's not like C++ is going anywhere anytime soon, (Some serious infrastructure reportedly still runs on COBOL after all!) but what if we could finally settle down on a technology for a certain kind of problem and focus all our energy on building new stuff? Wouldn't that be grand?
I wonder whether there's another industry that keeps reinventing itself every 20 to 30 years and still gets away with it.