Created attachment 136385 [details]
patch to fix RecentManager::add_item

-  c_data.groups = static_cast<gchar**>(g_malloc((data.groups.size() + 1) * sizeof(gchar)));
+  c_data.groups = static_cast<gchar**>(g_malloc((data.groups.size() + 1) * sizeof(*gchar)));


I asume that sizeof(*gchar) is the same as sizeof(gchar*), in which case this makes sense. This is likely to be a problem on 64-bit systems, right?

I think sizeof(*gchar) is just a syntax error, but sizeof(gchar*) is probably what was meant.  sizeof(gchar) is always 1 and definitely wrong in this context.

Alternatively, one may just use g_new().

sorry i made several typos, anyway i meant it should be sizeof(gchar*) and yes i have 64-bit system

Comment on attachment 136385 [details]
patch to fix RecentManager::add_item

>--- gtkmm-2.16.0/gtk/src/recentmanager.ccg	2009-06-12 04:32:58.000000000 +0100
>+++ gtkmm-2.16.0-patched/gtk/src/recentmanager.ccg	2009-06-12 05:28:12.000000000 +0100
>@@ -34,7 +34,7 @@ bool RecentManager::add_item(const Glib:
>   c_data.mime_type = const_cast<gchar*>(data.mime_type.c_str());
>   c_data.app_name = const_cast<gchar*>(data.app_name.c_str());
>   c_data.app_exec = const_cast<gchar*>(data.app_exec.c_str());
>-  c_data.groups = static_cast<gchar**>(g_malloc((data.groups.size() + 1) * sizeof(gchar)));
>+  c_data.groups = static_cast<gchar**>(g_malloc((data.groups.size() + 1) * sizeof(gchar*)));
>   for(unsigned int i = 0; i < data.groups.size(); ++ i)
>     c_data.groups[i] = const_cast<gchar*>(data.groups[i].c_str());
>   c_data.groups[data.groups.size()] = NULL;

Just to preempt confusion: This a bug on any machine and in no way limited to 64 bit systems.

On a side note, the code excerpt looks suspiciously like something that needs to be reworked anyway. Reasonable uses of g_malloc() in non-generated wrapper code are rare.

Applied in git Thanks. Please patch the ChangeLog in future.