It's not so much about tests being slow: some just get stuck and timeout
travis.
Skipped all tests taking about more than 0.2s to run on my laptop.
Fast testing takes about 8s instead of 24.
On Windows, the select.select() hangs/waits forever on the
test_async_connection_error_message() test. Adding a 10 second timeout
allows the tests to continue.
Postgres 9.3 turns messages about implicit indexes and sequences from NOTICE
to DEBUG1 so the tests fail with a default 9.3 server configuration because
the client doesn't get any NOTICE. Fix it by also asking for DEBUG1 messages
from the server when testing against Postgres >= 9.3.
If a connection is destroyed before an async operation is completed, the
`async_cursor` member creates a reference loop, leaving the connection and
the cursor alive. `async_cursor` is now a weak reference.
Dropped cyclic import from modules to tests: they were only working
because a second copy of the package was found in the project dir.
Use relative import so that 2to3 can do a good conversion.
If the connection is sync, notices will be processed by pq_fetch()
downstream.
If the connection is async, here we have only sent the query: no result
is ready yet, and neither notices have had a chance to arrive: they will
be retrieved later by pq_is_busy().
Added tests to check the above statement don't break.
Instead, the code should be using the fileno() and poll() methods of
the cursor's connection. Handle the case when poll() is called on an
already built connection as a request to poll the asynchronous query
(if there is one) and get NOTIFY events.
Update the tests to reflect that change, add a test for NOTIFY.
Do it by keeping the reference to the last PGresult in the cursor and
calling pq_fetch() before ending the asynchronous execution. This
takes care of handling the possible error state of the PGresult and
also allows the removal of the needsfetch flag, since now after
execution ends the results are already fetched and parsed.
This hides from the user the libpq's implementation detail of
requiring the first select() to wait for the connection socket to
become writable and makes it possible to have a uniform select loop
for both cursors and connections, in which you always start by polling
the object and then acting according to the result from poll().
Idea and implementation by Daniele Varrazzo.
If there is an asynchronous query, polling a cursor that did not
initiate it will raise an exception. Polling while there is no
asynchronous query underway still works, because the user needs to
have a way to get asynchronous NOTIFYs.
When a large query is sent to the backend (and probably in high
concurrency situations), writing the query could block. In
this case PQflush() should be called until it returns 0. The test checks
this is done correctly.
Some methods were forbidden in asynchronous mode, the isolation level
of an asynchronous connection is not always 0 and these changes
influenced expected test results.
