Code review comment for lp:~vanvugt/mir/mirout-save-load

> I wondered that myself. Why does MirBlob exist?...
>
> MirBlob has one benefit in that it is a single object with data and length
> fields. If MirBlob did not exist then the only real alternative is to pass
> around two variables instead:
> void *data;
> size_t size;
> Which is basically what MirBlob is.
>
> Your question however is also around data formats. That's a different question
> and the answer is that MirBlob is not a data format. The data format of the
> files here is from the Protobuf serialization function:
> https://developers.google.com/protocol-buffers/docs/reference/cpp/google.pr
> otobuf.message_lite#MessageLite.SerializeWithCachedSizes

Well, its more of compatibility of the file this utility produces.

If a user produces a file with this utility against mir 1.0.0
Then we change the internal structure of MirBlob in mir 1.x (as the internal structure isn't mirclient abi, this is something we could do)
Then the user tries to load the file with the mir 1.x utility, seems that this has a chance for breakage.