I commented in the jira about using int.
> On Aug 23, 2019, at 9:54 AM, Bruce Schuchardt wrote:
>
> Yeah, that code is confusing. I think the ordinal constants should all be
> changed to shorts and the constructor for Version should take a short as
> well. I can see why someone creating a ne
Yeah, that code is confusing. I think the ordinal constants should all
be changed to shorts and the constructor for Version should take a short
as well. I can see why someone creating a new ordinal would be hesitant
to do that - it will need to be tested.
The on-wire representation of a Vers
I think they did that because the datatypes in the Version class are too
small. I'm not sure why bytes and shorts were chosen.
The version constants are defined as bytes:
private static final *byte* GEODE_1_11_0_ORDINAL = 107;
While the ordinal is defined as a short:
private final *short* ordin
We've been incrementing the serialization version by 5 for each x.x.0
release but I see that we went from 105 in 1.10.0 to 107 in develop for
1.11.0. That gives us only one opportunity to make a serialization
change if we need to release patches for 1.10. I don't think that's
safe & we need t
