By binary format, you mean the serialized data.  In our scenario, shared 
memory is not used as a transport medium, it's the data itself.  Think of 
it as variables not shared between threads but between processes.  





Le mercredi 28 octobre 2020 à 17 h 38 min 29 s UTC-4, [email protected] a 
écrit :

> That could work but I suspect it would be a fair amount of work to 
> implement. I think at that point you would almost be implementing an 
> alternative serialization format for protocol buffers. Would it make sense 
> to just store the plain binary format in shared memory?
>
> On Wed, Oct 28, 2020 at 11:29 AM Mario Not64 <[email protected]> wrote:
>
>> Thanks for the feedback.
>>
>> Would it make sense for protobuf to create a class as per the .proto 
>> definition but with NO extra stuff, in its own .h file.   Then the real  
>> message proto class would derive from that.  Any code that used the class 
>> only need to include the .h of the class no need to drag all the protobuf 
>> stuff with it.  The message class would provide copy fonctions to easely 
>> transfer data from and to the real and message class. ?
>>
>> Le vendredi 23 octobre 2020 à 12 h 03 min 10 s UTC-4, [email protected] 
>> a écrit :
>>
>>> You are right that C++ protos are not POD and therefore might not be a 
>>> good fit for your use case. You could try looking at other protobuf 
>>> implementations such as UPB and protobuf-c. Since they are written in C I 
>>> think they meet the POD requirement.
>>>
>>> On Thu, Oct 22, 2020 at 8:56 PM Mario Not64 <[email protected]> wrote:
>>>
>>>> Hello,
>>>>
>>>> New to protobuf.  Like to implement it in our existing code base. We 
>>>> currently use cereal to serialiaze the data.  We have over 100 
>>>> structs/classes send over TCP/IP.  However reading about protobuf I came 
>>>> across an issue that I wounder if other have tackled and found a solution 
>>>> for.
>>>>
>>>> My understanding of protobuf is you defined your objects in .proto 
>>>> files, protobuf generates .h and .cc files (c++ in our case)  creating 
>>>> matching classes.  However most of are structs need to be POD (plain old 
>>>> data) because they also live in shared memory.  That seems totally 
>>>> incompatible with protobuf. That tells me when such struct need to be 
>>>> protobuffed , they must be copied into a protobuf version of that struct. 
>>>> That sounds like a high maintenancy, high risk proposition.
>>>>
>>>> Suggestion ?
>>>>
>>>> Regards,
>>>> Mario
>>>>
>>>> -- 
>>>> You received this message because you are subscribed to the Google 
>>>> Groups "Protocol Buffers" group.
>>>> To unsubscribe from this group and stop receiving emails from it, send 
>>>> an email to [email protected].
>>>> To view this discussion on the web visit 
>>>> https://groups.google.com/d/msgid/protobuf/de357190-edbb-442b-a952-39cec6a51814n%40googlegroups.com
>>>>  
>>>> <https://groups.google.com/d/msgid/protobuf/de357190-edbb-442b-a952-39cec6a51814n%40googlegroups.com?utm_medium=email&utm_source=footer>
>>>> .
>>>>
>>> -- 
>> You received this message because you are subscribed to the Google Groups 
>> "Protocol Buffers" group.
>> To unsubscribe from this group and stop receiving emails from it, send an 
>> email to [email protected].
>>
> To view this discussion on the web visit 
>> https://groups.google.com/d/msgid/protobuf/5497b8fb-2887-4ee6-8c13-ce705c2fa932n%40googlegroups.com
>>  
>> <https://groups.google.com/d/msgid/protobuf/5497b8fb-2887-4ee6-8c13-ce705c2fa932n%40googlegroups.com?utm_medium=email&utm_source=footer>
>> .
>>
>

-- 
You received this message because you are subscribed to the Google Groups 
"Protocol Buffers" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/protobuf/8181c7db-5a75-49c1-811a-8bb2ae614782n%40googlegroups.com.

Reply via email to