Tom

Yes, cases like that are what made me realize a `converters` function was worthwhile. In this case you'd use it like this (writing from memory) ```python def istrue(s): return s...

I see! The converter would indeed be complex in that case. Are you aware of a better option? In general, I think we should optimize interfaces for the common case,...

@flexatone thanks for the response! Storefilter works in the nan case, as shown in my example, but how would you use it to prevent `'None'`? I don't understand what causes...

Ok, I've been able to reproduce the `'None'` issue. This [tmp.xlsx](https://github.com/InvestmentSystems/static-frame/files/5191615/tmp.xlsx) is the same one as generated above, but in between writing it and loading it, I've opened and saved...

**Edit**: to insert the missing 'have' in the following sentence. But sometimes you have to specify a dtype conversion, e.g., to prevent an undesirable type guess. Perhaps the best solution...

That works, but if you *have* to specify a dtype, the `None` from openpyxl is converted to whatever dtype you've provided, and you have to replace that value instead of...

I think the approach of adding new interfaces is a fine one, but if I had my way we'd add `Frame.from_delimited_guess_types`, which would behave as `Frame.from_delimited` does now. We would...

Sounds reasonable!

Sounds good! I have made substantial progress, although I regret not finding time to finish. I'll open a PR with what I have, and comment more there.

In other cases, for which I haven't yet been able to produce a simple example, `nan` is similarly coerced to `False`.