Re: Can someone here shed some light on this ctypes/sqlite3 failure?
| Newsgroups | gmane.comp.python.db.pysqlite.user |
|---|---|
| Message-ID | <[email protected]> |
Gerhard> [email protected] wrote:
>> A number of us are having trouble with sqlite3 in Python in the
>> presence of ctypes on Macs. The information as I know it is here:
>>
>> https://sourceforge.net/support/tracker.php?aid=1581906
>>
>> Can anyone on this list shed some light on the problem?
Gerhard> I don't have a Mac, but after reading the messages, please try
Gerhard> this:
...
Thanks for the response. Mark Dickinson figured out the problem. When you
load any Apple library which relies on Apple's CoreData library you
implicitly load Apple's old SQLite library as well. Later on, even though
the sqlite3 extension module is linked to a newer version of libsqlite3 you
will still pick up some or all symbols from the old Apple-provided library.
I submitted a bug report to Apple though I don't expect any resolution from
them.
The workaround is (hopefully) to guarantee that the sqlite3 module is
imported before you begin mucking around with any of Apple's libraries which
depend on CoreData. For an application writer this may be feasible. For
Python I think we need to find a library to load in test_ctypes which
doesn't depend on CoreData. (I imagine Thomas picked GLUT because it is
widely available across platforms.)
Skip