C++, computer science, electronics, oxford commas, and running (he/him)
| Website | http://jsonlink.ca |
| Github | https://github.com/beached/ |
| https://twitter.com/beached_whale | |
| Website | https://o1fast.com/ |
JSON Link has C++26 reflection mostly baked in now. #cplusplus #cpp
Read Barry Revzin's blog [https://brevzin.github.io/c++/2024/09/30/annotations/] and it motivated me to update the C++26 Reflection support but with annotions to JSON Link. Code like
struct [[=daw::json::reflectable]] X {
[[=daw::json::rename_json{"member1"}]]
int m1;
[[=daw::json::rename_json{"member2"}]]
int m2;
};
Will (de)serialize to and from
{
"member1": 42,
"member2": 24
}
One of the things I like to do is compare how different languages solve the same problem — especially when they end up having very different approaches. It’s always educational. In this case, a bunch of us have been working hard on trying to get reflection — a really transformative language feature — into C++26. Fundamentally, reflection itself can be divided into two pieces:
Getting good bug reports and usage difficulty is so helpful in OSS libraries. New release of JSON Link today because of the help of others.
https://github.com/beached/daw_json_link/releases/tag/v3.25.0
The more I think about it, the more I think that #cplusplus OutputIterators are a mistake. At best they work ok with pointers, but that throws away size info. Outside of pointers, there is pretty much a forced size check on each call to operator* that may or may not be elided. It often isn't and types like back_inserter which calls push_back will almost never(not sure if it can) elide the check.
I think we would be much better off with a Writeable interface where the destination can do 1 size check for the full write instead of N of them.
TIL that I can open links to toots in my instance via https://<servername>/authorize_interaction?uri=<link to toot>
This is useful, without a dedicated plugin or with things like Alfred
C++ needs class/variable template specialization of things outside the current namespace e.g.
namespace ns {
struct Foo {};
template<>
inline constexpr bool ::other_ns::trait_v<Foo> = true;
}
It would make customizing things so much more ergonomic #cpp #cplusplus