How about using the binary format of the VM itself? Itâs a tested format that gives you instant compatibility with Pharo. I guess the VM people have documentation for it and could give you some pointers. -- Does this mail seem too brief? Sorry for that, I donât mean to be rude! Please see http://emailcharter.org <http://emailcharter.org/> . Johan Fabry - http://pleiad.cl/~jfabry PLEIAD and RyCh labs - Computer Science Department (DCC) - University of Chile
On 09 Nov 2016, at 17:27, Dimitris Chloupis <kilon.alios@gmail.com> wrote:
So now that CPPBridge at least basic are sort out I have to figure out a binary file format
The idea here is to implement an image format which will be part of the memory of the Pharo process but will live outside the memory of the Pharo VM. Lets call this CPPMemory will be stored to an image file, lets call it CPPImage, like we do with the Pharo image and will store a shared state among all other processes (Pharo or not) that connect to this shared memory (CPPMemory) via accessing the file (CPPImage). So basically it will act as an extension to the Pharo image.
In the image file , objects will be stored obviously not in a byte-code format but definitely in a byte format that will basically store the data of this object using very basic primitives like character , integers and of course bytes.
Obviously each object will have to have an id as a header and also its size in bytes and the size of each of its instance variables.
There will be also a global headers for the entire file with the list of the objects included and probably a references to other CPP images that they will be able to interconnect with each other like lego blocks.
Thats my vague idea, but I am open to suggestions and even links to articles and documentation about what a binary format must have to be considered well architectured. Obviously I dont expect perfection because this is my first binary file format but at least I am looking for advise for avoiding common pitfalls.