Better fail message

[Rlamboll]

Pull request

Please confirm that this pull request has done the following:

• Tests added
• Documentation added (where applicable)
• Example added (either to an existing notebook or as a new notebook, where applicable)
• Description in CHANGELOG.rst added

Adding to CHANGELOG.rst

Please add a single line in the changelog notes similar to one of the following:

- (`#XX <https://github.com/znicholls/silicone/pull/XX>`_) Added feature which does something
- (`#XX <https://github.com/znicholls/silicone/pull/XX>`_) Fixed bug identified in (`#YY <https://github.com/znicholls/silicone/issues/YY>`_)

[codecov]

Codecov Report

Merging #32 into master will increase coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master      #32      +/-   ##
==========================================
+ Coverage   99.23%   99.24%   +<.01%     
==========================================
  Files          12       12              
  Lines         394      397       +3     
  Branches       64       65       +1     
==========================================
+ Hits          391      394       +3     
  Misses          1        1              
  Partials        2        2

Impacted Files	Coverage Δ
...one/database_crunchers/quantile_rolling_windows.py	100% <100%> (ø)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update ad0b30b...c95e8f9. Read the comment docs.

[znicholls]

Why Warning rather than ValueError? An error would make more sense no? If users want to ignore it they can put it in a try, except block.

[Rlamboll]

We can do, although currently there are rather a lot of reasons why it may fail with a ValueError. This sort of warning will still need to be caught anyway. Do we want to ignore all ValueErrors from a single run of silicone in those cases? Quite possibly we do, I suppose.

[znicholls]

Do we want to ignore all ValueErrors from a single run of silicone in those cases?

So I think given this PR is just about what happens if there's no data, let's think about that first. As a user I'd expect an error with the message you have.

Then when we're running Silicone elsewhere (e.g. AR6 stuff) we can put the error handling in there. However in this PR I think an error is still most appropriate. If we wanted to get really fancy, we could create our own ValueError sub-class, NoData, or something.

[Rlamboll]

I've changed the error message, however I still get this weird tagging problem that means that half of the notepad checks fail. Am I just using a different package for something?

[znicholls]

Am I just using a different package for something?

No the order of columns in dataframes isn't stable. It's annoying but there it is. Just put # NBVAL_IGNORE_OUTPUT at the top of each of the failing cells.

[byersiiasa]

Thanks a lot @Rlamboll @znicholls - this is useful - so we will use a try/except block - but at least the warning will indicate why the particular scenario fails.

[znicholls]

Minor suggestion otherwise lgtm