-
InfluxDB
Power Real-Time Data Analytics at Scale. Get real-time insights from all types of time series data with InfluxDB. Ingest, query, and analyze billions of data points in real-time with unbounded cardinality.
-
SaaSHub
SaaSHub - Software Alternatives and Reviews. SaaSHub helps you find the best software and product alternatives
The prost serialize benchmarks are calling serialize\_pb() in the inner loop. That is creating a brand new instance of the message type. I see something similar in the capnp version here, but not for rkyv. I haven't used the rkyv API before so I may be misunderstanding, and they are all doing creation + encoding in the serialization inner loop?
I will also plug bytecheck, which I wrote specifically to address the security problems with rkyv. It allows you to thoroughly verify data integrity and I always recommend that people use it when dealing with data they cannot trust. For trusted data there's no issue, and there are other alternatives that are more performant for untrusted data as well.
Right, great question. I referenced one benchmark in the post that separated the populate and serialize steps. This is a bit of a fuzzy patch for the methodology: what should count as serialization? I think there could be arguments either way:
I think this is great. Could you add tree-buf as well to your benchmarks?
I encourage you to definitely try it out for your use case and see what works and what doesn't. If you do run into any sharp edges, add your experience to the feedback thread!
I don't think that rkyv will be able to guarantee set these bytes to zero (or any other value) after reading some internals discussions. You may be able to use the regular hash of your data to do something similar:
Related posts
-
A new major version of jql has been released
-
Pijul 1.0 Beta
-
Is there a library like Serde but which makes it easy to mutate serialized data stored in a [u8] or Vec<u8>?
-
rkyv 0.7: Endian-agnostic types, `no_std` validation, performance improvements, github sponsors and more!
-
rkyv 0.5: Comparison derives, serialize bounds, and the future