From 9e7500178fa2395c8955680e5c17c40a4b981407 Mon Sep 17 00:00:00 2001 From: Simon McVittie Date: Thu, 30 May 2013 17:01:16 +0100 Subject: nano version --- NEWS | 5 +++++ configure.ac | 2 +- 2 files changed, 6 insertions(+), 1 deletion(-) diff --git a/NEWS b/NEWS index dfb5b2fab..f3e8ff0b7 100644 --- a/NEWS +++ b/NEWS @@ -1,3 +1,8 @@ +telepathy-gabble 0.16.7 (UNRELEASED) +==================================== + +... + telepathy-gabble 0.16.6 (2013-05-30) ==================================== diff --git a/configure.ac b/configure.ac index a6ad9ed97..38b4a552e 100644 --- a/configure.ac +++ b/configure.ac @@ -10,7 +10,7 @@ AC_PREREQ([2.59]) m4_define([gabble_major_version], [0]) m4_define([gabble_minor_version], [16]) m4_define([gabble_micro_version], [6]) -m4_define([gabble_nano_version], [0]) +m4_define([gabble_nano_version], [1]) # Some magic m4_define([gabble_base_version], -- cgit v1.2.3 From 83bb468e0d4b386d88821ba2981f981214e9ebc6 Mon Sep 17 00:00:00 2001 From: Simon McVittie Date: Tue, 4 Jun 2013 12:40:45 +0100 Subject: Initialize libdbus for thread-safety libdbus is not thread-safe by default. This is a long-standing design flaw (). We call into GIO, which calls into glib-networking, which can (at least in recent versions) invoke libproxy in a thread. libproxy apparently has a Network-Manager plugin, which uses libdbus in that thread; meanwhile, we use libdbus in the main thread and everything goes badly for us. (It's possible that this crash is only reproducible with broken connectivity: I wrote this patch on a train, with intermittent mobile broadband coverage.) In libdbus < 1.7.4, libraries cannot safely initialize libdbus for multi-threading, because that initialization is not itself thread-safe (!); in particular, glib-networking cannot safely initialize libdbus. So, we have to do it. I have written patches to make libdbus thread-safe-by-default, but they haven't all been reviewed and merged yet, and in any case they won't be in a stable libdbus until 1.8. Until then, each application has to discover and fix this bug individually. Bug: https://bugs.freedesktop.org/show_bug.cgi?id=65296 Reviewed-by: Xavier Claessens --- src/gabble.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/src/gabble.c b/src/gabble.c index 0d2fb6e29..31ca34201 100644 --- a/src/gabble.c +++ b/src/gabble.c @@ -25,6 +25,8 @@ # include #endif +#include + #include #include @@ -116,6 +118,9 @@ gabble_init (void) g_thread_init (NULL); #endif + if (!dbus_threads_init_default ()) + g_error ("Unable to initialize libdbus thread-safety (out of memory?)"); + g_type_init (); wocky_init (); } -- cgit v1.2.3 From 5a349b2323f8bfd8a5d18c1bc2d79ef14689c0fe Mon Sep 17 00:00:00 2001 From: Simon McVittie Date: Tue, 4 Jun 2013 18:34:14 +0100 Subject: Disable unreliable test-case It has a race condition or something. Bug: https://bugs.freedesktop.org/show_bug.cgi?id=49595 Reviewed-by: Guillaume Desmottes --- tests/twisted/jingle-share/test-send-file-send-before-accept.py | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/tests/twisted/jingle-share/test-send-file-send-before-accept.py b/tests/twisted/jingle-share/test-send-file-send-before-accept.py index dd0a0287c..76b0c37ed 100644 --- a/tests/twisted/jingle-share/test-send-file-send-before-accept.py +++ b/tests/twisted/jingle-share/test-send-file-send-before-accept.py @@ -7,6 +7,10 @@ if not JINGLE_FILE_TRANSFER_ENABLED: print "NOTE: built with --disable-file-transfer or --disable-voip" raise SystemExit(77) +print("FIXME: test is not stable enough.\n" + + " https://bugs.freedesktop.org/show_bug.cgi?id=49595") +raise SystemExit(77) + class SendFileBeforeAccept(SendFileTest): def __init__(self, file, address_type, access_control, acces_control_param): -- cgit v1.2.3