Re: "ProgrammingError: can't adapt" when passing arguments to cursor.execute

Hi Federico,

I thought this problem might need you to solve. The problem is probably the
numpy.float32 . Can I fix this with a cast? Would float(...the thing that is
numpy.float32...) work?
Here's the type list for the script:
<type 'int'>
<type 'int'>
<type 'int'>
<type 'numpy.float32'>
<type 'str'>
<type 'bool'>
<type 'bool'>
<type 'bool'>
<type 'bool'>
<type 'bool'>
<type 'bool'>
<type 'bool'>
<type 'bool'>
<type 'bool'>
<type 'numpy.float32'>
<type 'numpy.float32'>
<type 'numpy.float32'>
<type 'numpy.float32'>
<type 'numpy.float32'>
<type 'str'>

Thanks,
Bill

On 11/20/09 11:36 AM, "Federico Di Gregorio" <fog-NGVKUo/i/[email protected]> wrote:

> Il giorno ven, 20/11/2009 alle 08.51 -0800, William Carithers ha scritto:
>> Thanks. I was worried about exactly that problem such as trying to load a
>> boolean into a column expecting a real. So I literally cut and pasted the
>> values from the "load" list in the script editor into the "load" list in the
>> interactive session. It worked in one case and not in the other. I'm not
>> sure why there is a roundoff difference but as long as values are consistent
>> about data types, I would think that it would be OK but at this point my
>> confidence level is pretty low.
> 
> Hi,
> 
> "can't adapt" has nothing to do with the destination types (the types of
> the backend columns). It is a Python problem: you passed to psycopg a
> type that psycopg does not know how to render into a string suitable to
> be passed to the backend.
> 
> Can you please add something like:
> 
> for x in load:
> print type(x)
> 
> to your code and check every parameter type?
> 
> federico