diff options
| author | Havoc Pennington <hp@redhat.com> | 2007-06-21 21:28:59 +0000 |
|---|---|---|
| committer | Havoc Pennington <hp@redhat.com> | 2007-06-21 21:28:59 +0000 |
| commit | ee721d13c455b5780af5989662ff7fca68374327 (patch) | |
| tree | ebcf018c66786c6588a4a2e9441904a401bbf26a | |
| parent | 5ff8975d9028749200a9e494255948a043743ee4 (diff) | |
2007-06-21 Havoc Pennington <hp@redhat.com>
* dbus/dbus-watch.c (dbus_watch_get_fd): 1) its behavior should
not be the same as before, the ABI has never been declared stable
on Windows and 2) do not commit to cross-platform files without
posting the exact patch to the mailing list
* dbus/dbus-sysdeps-util.c (_dbus_sysdeps_test): as I have pointed
out before, either 0xff needs to work on both platforms, or none
of the dbus code can rely on it working. That means the options
are 1) audit the code for anywhere that relies on 0xff working, if
none found ideally add a test that it *doesn't* work and make unix
reject it explicitly, but in any case this test would go away or
2) make it work on Windows also, then we don't have to figure out
whether we rely on it.
And in either case, post the exact patch to the mailing list and
don't just commit.
| -rw-r--r-- | ChangeLog | 19 | ||||
| -rw-r--r-- | dbus/dbus-sysdeps-util.c | 3 | ||||
| -rw-r--r-- | dbus/dbus-watch.c | 4 |
3 files changed, 20 insertions, 6 deletions
@@ -1,3 +1,22 @@ +2007-06-21 Havoc Pennington <hp@redhat.com> + + * dbus/dbus-watch.c (dbus_watch_get_fd): 1) its behavior should + not be the same as before, the ABI has never been declared stable + on Windows and 2) do not commit to cross-platform files without + posting the exact patch to the mailing list + + * dbus/dbus-sysdeps-util.c (_dbus_sysdeps_test): as I have pointed + out before, either 0xff needs to work on both platforms, or none + of the dbus code can rely on it working. That means the options + are 1) audit the code for anywhere that relies on 0xff working, if + none found ideally add a test that it *doesn't* work and make unix + reject it explicitly, but in any case this test would go away or + 2) make it work on Windows also, then we don't have to figure out + whether we rely on it. + + And in either case, post the exact patch to the mailing list and + don't just commit. + 2007-06-21 Ralf Habacker <ralf.habacker@freenet.de> * cmake/CMakeLists.txt: added VERSION_PATCH to be able to diff --git a/dbus/dbus-sysdeps-util.c b/dbus/dbus-sysdeps-util.c index c5aecb95..1f6ceb9b 100644 --- a/dbus/dbus-sysdeps-util.c +++ b/dbus/dbus-sysdeps-util.c @@ -141,7 +141,6 @@ _dbus_sysdeps_test (void) exit (1); } -#ifndef DBUS_WIN _dbus_string_init_const (&str, "0xff"); if (!_dbus_string_parse_double (&str, 0, &val, &pos)) @@ -159,7 +158,7 @@ _dbus_sysdeps_test (void) _dbus_warn ("_dbus_string_parse_double of \"0xff\" returned wrong position %d", pos); exit (1); } -#endif + #ifdef DBUS_WIN check_path_absolute ("c:/", TRUE); check_path_absolute ("c:/foo", TRUE); diff --git a/dbus/dbus-watch.c b/dbus/dbus-watch.c index 634ab247..eed9a105 100644 --- a/dbus/dbus-watch.c +++ b/dbus/dbus-watch.c @@ -489,11 +489,7 @@ _dbus_watch_set_handler (DBusWatch *watch, int dbus_watch_get_fd (DBusWatch *watch) { -#ifdef DBUS_WIN_FIXME - return watch->fd; -#else return dbus_watch_get_unix_fd(watch); -#endif } /** |