6 commits

Author	SHA1	Message	Date
Thomas Haller	a6be2f4aa9	all: use nm_str_hash() instead of g_str_hash() ... We also do this for libnm and libnm-core, where it causes visible changes in behavior. But if somebody would rely on the hashing implementation for hash tables, it would be seriously flawed.	2017-11-16 11:49:52 +01:00
Beniamino Galvani	ade90756ba	libnm-core: add ifcfg-rh documentation for user setting	2017-09-05 10:33:42 +02:00
Thomas Haller	488029d74b	libnm: use enum for setting priorities	2017-06-07 09:07:17 +02:00
Thomas Haller	c429951c46	libnm: track invalid user data separately and reject during verify() ... nm_setting_user_set_data() rejects invalid keys and values, and can fail. This API is correct never to fail, like the get_data() only returns valid user-data. However, the g_object_set() API allows to set the hash directly but it cannot report errors for invalid values. This API is used to initialize the value from D-Bus or keyfile, hence it is wrong to emit g_critial() assertions for untrusted data. It would also be wrong to silently drop all invalid date, because then the user cannot get an error message to understand what happend. The correct but cumbersome solution is to remember the invalid values separately, so that verify() can report the setting as invalid. (cherry picked from commit 1dbbf6fb03)	2017-05-06 14:53:05 +02:00
Beniamino Galvani	1d08391ed1	all: fix typo 'entires' -> 'entries' ... (cherry picked from commit a4cbfe6f0a)	2017-05-05 19:16:39 +02:00
Thomas Haller	4ec7dd987e	libnm: add NMSettingUser ... This only adds new API for a NMSettingUser. The setting class is still entirely unused. The point is getting the new API into 1.8.0 release of libnm. It's easier to backport the use of the API to a stable branch then backporting public API. https://bugzilla.gnome.org/show_bug.cgi?id=776276 https://bugzilla.redhat.com/show_bug.cgi?id=1421429	2017-03-28 14:58:21 +02:00