[SciPy-Dev] requesting code review: added close matches to constants.codata.value()

[Robert Kern]

On Mon, Apr 28, 2014 at 1:48 PM, Pierre Haessig
<pierre.haessig at crans.org> wrote:
> Hi Kevin,
>
> Just a random thought:  what about putting your recommendation system in
> a separate "search" function ?
> The KeyError raised by value(...) could be one-line message pointing to
> this search function.

The argument to KeyError should be the key that is causing the
problem, not a natural language message. That's why his attempt to use
a multi-line string got printed as its repr().

> That function, meant to be used interactively unlike value(), could use
> prints I guess.

But yes, I agree that adding a function specifically documented to be
used interactively would be a better place for this functionality.

-- 
Robert Kern