python shell handling is broken.

[linas]

See comments in pull request opencog/atomspace#838

At this time, starting the cogserver, and then telnet localhost 17001, entering the python shell, and entering this:

for i in range(10) :
    print "foo", i

produces no output in either the shell or the cogserver (see also issue #2 -- currently prints go to the cogserver; once fixed, prints should go to the shell.) There are also various inconsistencies: running the python shell stand-alone gives this:

$ python
Python 2.7.6 (default, Jun 22 2015, 17:58:13) 
[GCC 4.8.2] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> 
>>> 
>>> 
>>> 

after hitting carriage return 3 times. Doing the same in the telnet terminal gives this:

$ telnet localhost 17001
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.
opencog> py
Entering python shell; use ^D or a single . on a line by itself to exit.
py> 
... 
... 
py> 
py> 
... 
... 
... 
... 
... 
... 
py> 
... 
... 
... 
... 
py> 
... 

There are also inconsistencies in what python considers to be "legal syntax". For example, the following code, when entered at the python shell, is invalid syntax:

def fun(x):
    y = x + 1

    print "foo", y
    return y

fun(1)
print "bar"

it generates this error:

$ python
Python 2.7.6 (default, Jun 22 2015, 17:58:13) 
[GCC 4.8.2] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> def fun(x):
...     y = x + 1
... 
>>>     print "foo", y
  File "<stdin>", line 1
    print "foo", y
    ^
IndentationError: unexpected indent
>>> 

however, copying that code to a file, say f.py and running it as $ python f.py does not produce any error:

$ python f.py
foo 2
bar
$

running the above code at the telnet prompt does nothing.

I'm not sure quite how to resolve this ... if python itself is not self-consistent with what it considers to be valid syntax, I'm not sure how how to emulate that .. at any rate, the telnet shell is clearly broken.

[linas]

One reason the telnet shell is broken is that we have no unit tests for it. Once upon a time, it used to work.

[ngeiswei]

python results are discarded so it's difficult to test it. Agreed it is broken, but I guess we'll address that more deeply after the demo.

[smigad]

I sort of got it working
Code like this works fine

for i in range(10) :
    print "foo", i

but code like the one u posted with the empty line inside a scope doesn't work
because like the standard interpreter, an empty line ends a scope(when typed in).
https://github.com/Dagiopia/atomspace/blob/master/opencog/cython/PythonEval.cc

[linas]

I stubbed out line 107 of tests/cython/PythonModuleUTest.cxxtest until this issue is resolved. When this bug is fixed, then that line should be enabled again.