DataReader should cache queries.

[femtotrader]

Hello,

each request using pandas.io.data.DataReader ends by a server request.
That's not reasonable.
Maybe a cache for DataReader should be considered.

Kind regards

[jreback]

see here: #6456

I think this is out-of-scope for pandas directly. That said if someone wanted to implemented this (via the requests library, I think we would accept it).

@jorisvandenbossche
@TomAugspurger
@cpcloud

?

[TomAugspurger]

Agreed with it being out of scope.

[MichaelWS]

requests_cache is relatively transparent, but I am not sure on how unreasonable it is.

If I have a function that gets historical data from yahoo, I feel like the caching should be done by choice by the user.

[femtotrader]

Hello,

I've been working on this issue see
https://github.com/femtotrader/pandas_datareaders
but I need some help with some DataReaders.

[jreback]

@kay1793 let try to have constructive comments
thank you for all you contributions

[kay1793]

@jreback, I'm trying but you censoring criticism and erasing comments isn't helping.
Would you please explain what you mean by it is out of scope but will accept PR? I still find this confusing.

[MichaelWS]

I will step in. This is a discussion about separating pandas datareader functionality from pandas.

Pandas is intended to be a middleware data layer.

[davidastephens]

This was added to pandas-datareader v0.2.1, so this issue can be closed.