delta/apache/apr.git/test, branch pollcb-dev

Add a test for a timeout, but with a socket recently read from inside the pollcb.

2006-03-31T05:35:52+00:00

git-svn-id: https://svn.apache.org/repos/asf/apr/apr/branches/pollcb-dev@390332 13f79535-47bb-0310-9956-ffa450edef68

Add a new timeout testcase, which makes sure that with no sockets inside the pollset, we should trigger a timeout.

2006-03-31T05:26:50+00:00

git-svn-id: https://svn.apache.org/repos/asf/apr/apr/branches/pollcb-dev@390330 13f79535-47bb-0310-9956-ffa450edef68

Add a simple Callback testcase, to make sure it kind-of-works. YMMV :)

2006-03-31T05:21:32+00:00

git-svn-id: https://svn.apache.org/repos/asf/apr/apr/branches/pollcb-dev@390328 13f79535-47bb-0310-9956-ffa450edef68

Add a first revision of the Pollset Callback API.

2006-03-31T05:07:04+00:00

Currently it only has an EPoll implementation, and the most basic test case possible.


git-svn-id: https://svn.apache.org/repos/asf/apr/apr/branches/pollcb-dev@390326 13f79535-47bb-0310-9956-ffa450edef68

* test/testsockets.c

2006-03-22T21:22:49+00:00

  (sendto_receivefrom): Bail out if apr_socket_bind fails, as has been
   known to happen on solaris if you turn on ipv6.


git-svn-id: https://svn.apache.org/repos/asf/apr/apr/trunk@387950 13f79535-47bb-0310-9956-ffa450edef68

Fix a longstanding missunderstanding in our kqueue pollset code. It turns

2006-03-15T19:59:51+00:00

out that the kqueue filter types are not bitfields, so checking for them
via & EVFILT_READ or & EVFILT_WRITE is inappropriate.

Test Fixes By: Joe Orton

* poll/unix/kqueue.c
  (get_kqueue_revent): Use == instead of & when testing for filter types.

* test/testpoll.c
  (multi_event_pollset): Handle the fact that we can sometimes get multiple
   events for a single socket.

* CHANGES: Note fix.


git-svn-id: https://svn.apache.org/repos/asf/apr/apr/trunk@386154 13f79535-47bb-0310-9956-ffa450edef68

There are -so- many reasons this code -could- fail, and any

2006-03-13T16:46:18+00:00

  of them will hang the remainder of the test.  Better to simply
  fail them all, with an extra notice when the error is unusual.

git-svn-id: https://svn.apache.org/repos/asf/apr/apr/trunk@385594 13f79535-47bb-0310-9956-ffa450edef68

* configure.in: Set LT_NO_INSTALL to -no-install on all platforms but

2006-03-10T10:09:36+00:00

Darwin.

* test/Makefile.in: Revert r384715 and use LT_NO_INSTALL instead of
hard-coding -no-install.


git-svn-id: https://svn.apache.org/repos/asf/apr/apr/trunk@384750 13f79535-47bb-0310-9956-ffa450edef68

Throw away a horrible change; each time we layer on more garbage to libtool

2006-03-10T06:40:52+00:00

  et al we've caused apr to fail in more peculiar ways on more platforms.

  In this case, let libtool do exactly what we want, which is to resolve the
  build-path flavor of libapr-1.la and test exactly what we've most recently
  changed (and, for that matter, to have make check function whatsoever before
  the library has been installed.)


git-svn-id: https://svn.apache.org/repos/asf/apr/apr/trunk@384715 13f79535-47bb-0310-9956-ffa450edef68

* test/testnames.c (root_from_cwd_and_back): Fix segfault on

2006-03-01T14:33:03+00:00

non-Win32.


git-svn-id: https://svn.apache.org/repos/asf/apr/apr/trunk@382033 13f79535-47bb-0310-9956-ffa450edef68