[python-win32] datetime values in adodbapi

[Bob Kline]

The adodbapi package does not seem to handle datetime values appropriately.

import adodbapi
print(adodbapi.version)
cursor = adodbapi.connect(...)
cursor.execute("CREATE TABLE #t (i INT, d DATETIME)")
cursor.execute("INSERT INTO #t VALUES (42, GETDATE())")
cursor.execute("SELECT * FROM #t")
row = cursor.fetchone()
print("d={}".format(row.d))
cursor.execute("INSERT INTO #t VALUES (?, ?)", (43, row.d))
cursor.execute("SELECT * FROM #t WHERE i = 43")
row = cursor.fetchone()
print("d={}".format(row.d))

When running under Python 2.7.13, the first print statement shows that
the datetime value has been correctly retrieved, preserving the
precision:

d=2017-11-04 07:52:18.110000

but then when the second INSERT is executed an exception is raised:

pywintypes.com_error: (-2147352567, 'Exception occurred.', (0,
u'ADODB.Parameter', u'Application uses a value of the wrong type for
the current operation.', u'C:\\WINDOWS\\HELP\\ADO270.CHM', 1240652,
-2146824867), None)

So the package isn't allowing the script to use the same value it was
given as a valid value for the d column by the package itself.

When the script is run under Python 3.6.0, the exception is not
raised, but that's only because the datetime precision has been
discarded:

d=2017-11-04 08:15:37
d=2017-11-04 08:15:37

I get the same behavior with adodbapi 2.6.0.6 (which I downloaded from
the link which indicated (http://adodbapi.sourceforge.net/index.html)
that this is the latest version) and on 2.6.0.7, which is what pip
installed.

I would like to help get both of these behaviors corrected (failure to
store datetime values with microsecond precision in Python 2.x, as
well as failure to preserve sub-second precision in retrieved values
in Python 3.x). Before I dig in to do the work, though, I want to
check in here to make sure patches are accepted so I don't spend time
on something that's not wanted.

Thanks!

-- 
Bob Kline
http://www.rksystems.com
mailto:bkline at rksystems.com