Can I see an example of a ndarray data set having missing values.
By convention is the missing values represented as None or isnan ?

@MichaelArthurAnderson your views?

I have mentioned in the climate issues

Thank you @prateekiiest
Over on Jira this issue was reported by @huikyole, he is on vacation right now but I am sure he can answer your question when he returns.

@prateekiiest once we get an example of ndarray data set having missing values it would be great tp create a unit test for this.

Thank you @lewismc for letting me know about it. I look forward to it.

Tests seem to fail in case of dependencies netcdf4

@prateekiiest do you mean it seems to fail? Can you elaborate?

Oops, sorry I actually meant to say fail. Maybe slipped off.

Actually Travis test fails for py2.7 and py3.5

May be some dependency issues?

I saw a PR recently related to netcdf4. #507

I am not sure if its related to that.

@huikyole @agoodm can you guys please scope out this PR. Thanks, we are working on the build issues.

Ah, I take it this is why you added cython to the dependencies list @lewismc ? conda has been having all sorts of issues with dependency trees lately, as recently @huikyole and I both discovered that pyyaml wasn't actually being included in RCMES distributions like it was supposed to be. We had to add that dependency explicitly to our builds even though we didn't need to before.

Oh, and please close and reopen the PR @prateekiiest so we can see the build run again.

ok, @agoodm

@lewismc still fails for pydap as in the other PR #503

Hi @prateekiiest please close and reopen again, I added netcdftime to the build profiles

@huikyole could you please look into the present approach?