Code review comment for lp:~mir-team/mir/alternative-extension-naming

> Alright, so there's human-readable strings now.
>
> The remaining question (which seems slight) is how to give the user a
> connection between a struct containing the functions and the number to plug in
> as the version to get the appropriate struct.
>
> The current contenders seem to be:
>
> 1) struct-with-version-in-name
> struct MirExtensionFooV1 (and then "1" is the int to plug in as the version)
> 2) existing, just have a #define
> 3) Alan's pseudocode (not quite clear to me how this would look in C code, so
> can't comment)
>
> I don't mind 1) really, but it seems less advantageous to me because of the
> mental hop that is needed (mentally parse "V1", and then use it as a version).
>
> Will wait to change the code until consensus, the differences are slight (or a
> matter of preference, perhaps).

I vote for (2). (2) is more explicit to me, whereas (1) is more implicit and could be confusing. People are used to khronos style extension strings already.