FlatBuffers VS cereal

According to the official website:

> Protobuf is intentionally designed to NOT require any parsing at all.

As others have mentioned, this is simply not the case, and the VARINT encoding is a trivial counterexample.

It is this required decoding/parsing that (largely) distinguishes protobuf from Google's flatbuffers:

https://github.com/google/flatbuffers

https://flatbuffers.dev/

Cap'n Proto (developed by Kenton Varda, the former Google engineer who, while at Google, re-wrote/refactored Google's protobuf to later open source it as the library we all know today) is another example of zero-copy (de)serialization.

I don't work at Cloudflare but follow their work and occasionally work on performance sensitive projects.

If I had to guess, they looked at the landscape a bit like I do and regarded Cap'n Proto, flatbuffers, SBE, etc. as being in one category apart from other data formats like Avro, protobuf, and the like.

So once you're committed to record'ish shaped (rather than columnar like Parquet) data that has an upfront parse time of zero (nominally, there could be marshalling if you transmogrify the field values on read), the list gets pretty short.

https://capnproto.org/news/2014-06-17-capnproto-flatbuffers-... goes into some of the trade-offs here.

Cap'n Proto was originally made for https://sandstorm.io/. That work (which Kenton has presumably done at Cloudflare since he's been employed there) eventually turned into Cloudflare workers.

Another consideration: https://github.com/google/flatbuffers/issues/2#issuecomment-...

Much more in the discussion of this particular issue onGitHub: flatbuffers:Reverse engineering #4258

Related but not Rust-specific: FlatBuffers, Cap'n Proto.

If json files are a concern because of space, you can always look into something like protobuffers or flatbuffers. But whatever you use, you should try to find a solution where you don't have to think about the actual serialization/deserialization of your objects, and can just concentrate on the data.

Or, as an alternative to protobuffers, there's also flatbuffers, which is lighter weight and needs less memory: https://google.github.io/flatbuffers/

I'd stick with cereal for serialization

Today I'm back to working on Pataro, the roguelike built on libtcod that made up much of my Hacktoberfest efforts. I had been assigned to an issue requesting the addition of serialization and deserialization, but unfortunately ran out of time and wasn't able to finish the former or start the latter. I ran into issues with Cereal, and had a hard time figuring out the structure of the program and how to go about implementing serialization for all the relevant components. At the end of that attempt I mentioned that if I were to try again I'd start by testing out Cereal separately and getting a handle on that before trying to implement it in Pataro - so that's what I'm doing today.

I am once again working on Pataro today, and I've succeeded in clearing up some issues and creating new ones. I've been stuck on an issue where Visual Studio was raising errors in the portion of the code where I call the cereal archive on some types, but was able to clear up that issue by moving the save function definitions out of their respective headers and into the corresponding .cpp files. Examining this repo and its use of cereal again, I was able to get a bit clearer of an idea of how it's implemented, and I included just about every relevant cereal header I could find to try and avoid any issues like the previous one popping up again (with the intention of later removing whichever I can to avoid redundant inclusions).

On the Pataro front, I've started looking at other examples of people using cereal for their games. It seems to be a popular choice for roguelike games like this, so hopefully I can figure out both the syntax problems I'm facing, as well as logical ones like how the program should be structured to have all the necessary data properly serialized and deserialized.

Progress on one front but roadblocks on another. After my post yesterday outlining my plans to add serialization to Pataro I ran into some issues with calling the archive class in cereal wherein the call wouldn't go through due to an incorrect number of arguments. This happened with both vector and size_t data members, and I'm still investigating why.

This is the simplest serialization library I saw. You only need to add one method to your class and you can stream it to desired archive. It supports binary as well as json, XML and others. https://github.com/USCiLab/cereal