Merge lp:~mbp/launchpad/mbp-trivial into lp:launchpad

[Martin Pool]

handle EINTR in test_on_merge so resizing the window doesn't crash your test run

Same as https://code.edge.launchpad.net/~mbp/launchpad/mbp-trivial/+merge/32173 but into devel

[Michael Hudson-Doyle (mwhudson)]

Looks fine, thanks for making the comment clearer.

review: Approve

[Martin Pool (mbp)]

It fails like so:

Tests with errors
-----------------
 lp.code.windmill.tests.test_branch_subscriptions.TestBranchSubscriptions.test_branch_subscription_ajax_load
 lp.codehosting.puller.tests.test_worker.TestWorkerProgressReporting.test_network

======================================================================
ERROR: lp.code.windmill.tests.test_branch_subscriptions.TestBranchSubscriptions.test_branch_subscription_ajax_load (subunit.RemotedTestCase)
----------------------------------------------------------------------
_StringException: Text attachment: garbage
------------
[<Thread(Thread-631, started daemon 47931595147024)>]
------------

======================================================================
ERROR: lp.codehosting.puller.tests.test_worker.TestWorkerProgressReporting.test_network (subunit.RemotedTestCase)
----------------------------------------------------------------------
_StringException: Text attachment: garbage
------------
[<Thread(Thread-18, started daemon 47106544203536)>]
------------

**NOT** submitted to PQM:
[r=mwhudson][ui=none][bug=615740] handle EINTR in test_on_merge so resizing the window doesn't crash your test run

I guess this is not my fault, but I don't know.

[Michael Hudson-Doyle (mwhudson)]

The branch passed on EC2 a second time, but then failed to land due to testfix mode. Can you land this yourself?

[Martin Pool (mbp)]

On 25 September 2010 07:26, Michael Hudson-Doyle
<email address hidden> wrote:
> The branch passed on EC2 a second time, but then failed to land due to testfix mode.  Can you land this yourself?

All my previous attempts have been rejected, but I can see about
having pqm access.

--
Martin

=== modified file 'test_on_merge.py'
--- test_on_merge.py	2010-08-10 21:27:56 +0000
+++ test_on_merge.py	2010-09-23 23:57:49 +0000
@@ -1,6 +1,6 @@
 #!/usr/bin/python -S
 #
-# Copyright 2009 Canonical Ltd.  This software is licensed under the
+# Copyright 2009, 2010 Canonical Ltd.  This software is licensed under the
 # GNU Affero General Public License version 3 (see the file LICENSE).
 
 """Tests that get run automatically on a merge."""
@@ -13,7 +13,7 @@
 import psycopg2
 from subprocess import Popen, PIPE, STDOUT
 from signal import SIGKILL, SIGTERM, SIGINT, SIGHUP
-from select import select
+import select
 
 
 # The TIMEOUT setting (expressed in seconds) affects how long a test will run
@@ -164,7 +164,24 @@
     # Popen.communicate() with large data sets.
     open_readers = set([xvfb_proc.stdout])
     while open_readers:
-        rlist, wlist, xlist = select(open_readers, [], [], TIMEOUT)
+        # select() blocks for a long time and can easily fail with EINTR
+        # <https://bugs.launchpad.net/launchpad/+bug/615740>.  Really we
+        # should have EINTR protection across the whole script (other syscalls
+        # might be interrupted) but this is the longest and most likely to
+        # hit, and doing it perfectly in python has proved to be quite hard in
+        # bzr. -- mbp 20100924
+        while True:
+            try:
+                rlist, wlist, xlist = select.select(open_readers, [], [], TIMEOUT)
+                break
+            except select.error, e:
+                # nb: select.error doesn't expose a named 'errno' attribute,
+                # at least in python 2.6.5; see
+                # <http://mail.python.org/pipermail/python-dev/2000-October/009671.html>
+                if e[0] == errno.EINTR:
+                    continue
+                else:
+                    raise
 
         if len(rlist) == 0:
             # The select() statement timed out!