-
selsta.merge+ 10648 10649
-
xmr-prAdded
-
br-m<jpk68:matrix.org> vtnerd: CDDL seems like a really cool idea
-
vtnerdyes, its easier to read than json-schema. the only real drawbacks of cbor are listed too, but I may come around to accepting htem
-
br-m<jpk68:matrix.org> I also feel like documentation for CBOR RPC will be substantially easier, considering its standarization and industry adoption
-
br-m<jpk68:matrix.org> Pretty sure Intel has their own CBOR library on GH, which might have SIMD support, I can't remember
-
vtnerdI don't see SIMD support would be helpful, there's no whitespace and you can't read multiple integers at a time due to the tag design
-
vtnerdcbor-rpc will not be substantially easier, except for maybe grokking cddl vs json-schema (which is what I'd use with msgpack)
-
vtnerdlike in either case you've gotta transform bytes into data structures via some lib or custom code
-
vtnerdand both are well-defined. cbor is just more annoying with float16 and indefinite length items
-
vtnerdotherwise its practically the same thing
-
selstagingeropolous: you might have an opinion regarding this, as it was your idea initially monero-project/monero #10690
-
vtnerdok, I just put together a cbor encoder in less than 3 hours, and its seriously bad-ass from the encoding side
-
vtnerdall signed/unsigned output can merge into one function easily, unlike msgpack
-
vtnerdthe cache locality has to be better, but I'd have to do a valgrind run to verify (as my ryzen chip just crushes it)
-
br-m<syntheticbird> CBOR SUPPORT INCOMING IN MONEROD 2026 CONFIRMED
-
vtnerdthe size between msgpack is really similar, and so is the timing (but integer keys are actually really close to viable)
-
vtnerdlol are you excited too?
-
br-m<syntheticbird> of course i am
-
br-m<jpk68:matrix.org> CBOR FTW