TorbenCK

Good to hear. Yes, we currently thinking about if we fix this for ourself and then submit it as PR here.

Looks good for me 👍 I like the Testcase in [7d7ab59](https://github.com/liip/sheriff/pull/43/commits/7d7ab59c5ebbe0bf162f5126dcc4f77b24e41940) where json.Marshal is compared with sheriff.Marshal. Maybe we could extend this kind of testcase also for pointers. But currently...

Yes, the change might be small, but might not be minor. It would lead to a regression, if a client expects that arrays are never null. It is the same...

Another solution would be to make sheriff configurable. By default it will have the old behaviour, but the json.Marshal conform behaviour can be activated over configuration. This should also cover...

Just wanted to follow-up as we are interested in a near-term solution for the problem. Can we release https://github.com/liip/sheriff/pull/43 soon?

just wanted to follow-up. I am not sure what still needs to be done but hope someone will find time soon, so we can utilize this lib in our solution...