messages flash size are really big for an embedded

[farbodpm]

I used protocol buffer in C. These days I am trying to use my codes in rust so I am developing code using ptost. Recently I have find out that prost mesages are really big in flash size. I want to take advice about how to use prost or how to optimize my code.

[LucioFranco]

You may want to consider using more embedded focused proocols, I don't work much in the embedded space so don't really know the requirements that well but I am sure prost doesn't do much to help that use case that mucn.

[farbodpm]

Dear @LucioFranco ,
Thanks for your response. I have experience in using protobuf in c and that was okay. Now we are migrating from c to rust so that these problems showed up. I was checking prost library last nigh and I realized that the Error handling are so much and so detailed. That's great but a little annoying for embedded systems. May be by adding some optimization the code get much smaller.
It would be my pleasure to help.

[farbodpm]

I also have q question. In a message like below. Ho does the trait of encoding and merge would be created for "all" message?
message A{
uint32 a = 1;
}
message B{
uint32 a = 1;
}
message C{
uint32 a = 1;
}
message all{
repeated A a =1;
repeated B b =2;
repeated C c =4;
}

[LucioFranco]

Im not sure what you mean by merge here? Could you explain more?

[ncs-pl]

Your problem actually reflects the lack of support for options in Prost!
I think your problems (the main message as well as your question on duplicate generation) could be solved by supporting the optimize_for file option.

The CODE_SIZE or LITE_RUNTIME optimizations would make the code generation more efficient (with, for example, a Reflection-based generation when similar fields are found in various structs).

There are currently open pull requests trying to implement the support for options/custom extensions (both being extremely linked). As for now, you should consider keeping your communication protocols based on Protocol Buffers on C/C++ codes.