While converting our application to use pg.DB().query_formatted() ..
I discovered that pg.DB() leaks, while pg.connect() doesn't. Here's a minimal
test case. I'm sorry, but development is not my job and python not my
expertise so I have to leave it here...
Look for "maximum resident set size" in output of:
command time -v python ./pgleak.py
#!/usr/bin/python
import pg
import thread
_db={}
id=thread.get_ident()
def query(text):
try:
return _db[id].query(text)
except KeyError:
pass
_db[id]=pg.DB('ts')
return _db[id].query(text)
def close():
#print 'l',len(_db)
_db.pop(id).close()
while 1:
res = query('SELECT 1 FROM sites').dictresult()
close()
_______________________________________________
PyGreSQL mailing list
[email protected]
https://mail.vex.net/mailman/listinfo.cgi/pygresql
