[C++-sig] Re: [BPL] python::interpreter proposal

[Dirk Gerrits]

Bjorn Pettersen wrote:
>>From: Dirk Gerrits [mailto:dirk at gerrits.homeip.net] 
>>
>>Bjorn Pettersen wrote:
>>
>>>>From: Dirk Gerrits [mailto:dirk at gerrits.homeip.net]
>>>>
>>>>For the python::interpreter class that has been discussed
>>>>here earlier I propose the following interface:
>>>
>>>[...]
>>>
>>>
>>>>object interpreter::run(object code_string);
>>>>
>>>>Run the Python code in the string
>>>>extract<std::string>(code_string) with 
>>>>this (sub-)interpreter and return the result.
>>>
>>>
>>>Question... how are you going to distinguish calling PyRun_String
> 
> with 
> 
>>>Py_eval_input vs. Py_file_input? (the former only takes expressions,
> 
> 
>>>the latter doesn't return a result...)
>>
>>It doesn't? I wasn't aware of that.
> 
> 
> No, it's undocumented.
> 
> 
>>The PyRun_String docs say: "Returns the result of executing 
>>the code as a Python object, or NULL if an exception was raised."
> 
> [...]
> 
> Try e.g. 
> 
>   PyObject* res = PyRun_String("5", Py_file_input, ns, ns);
> 
> and you'll see that res is pointing to None afterwards... Quite annoying
> :-)

Hmm that's quite a showstopper. Perhaps a second enum parameter is 
needed? Something like:

enum code_type
{ single_expression, single_statement, multiple_statements };

Not very nice IMHO so if anyone knows a way around it, please enlighten 
me. ;)

Dirk Gerrits