GUdevClient always creates a monitor instance, even if there are no subsystems
or handlers defined. Hence the first iteration of NMUdevClient did that as
well.
I think that can be avoided however. We only need a monitor when there is
a event handler subscribed. Contrary to GUdevClient, we know that from the
very beginning.
_NM_GET_PRIVATE() macro is used to implement a standard private-getter, but it
requires that "self" is a pointer of either "const type *" or "type *". That
is great in most cases, but sometimes we have predominatly self pointers of
different type, so it would require a lot of casts.
Add a different form _NM_GET_PRIVATE_VOID() where self pointer can be any
non-const pointer and returns a non-const private pointer after casting.
I think NM_CACHED_QUARK_FCN() is better because:
- the implementation is in our hand, meaning it is clear that
putting a "static" before NM_CACHED_QUARK_FCN() is guaranteed to
work -- without relying on G_DEFINE_QUARK() to be defined in a way
that this works (in fact, we currently never do that and instead
make all functions non-static).
- it does not construct function names by appending "_quark".
Thus you can grep for the entire function name and finding
the place where it is implemented.
- same with the stings, where the new macro doesn't stringify the
argument, which is less surpising. Again, now you can grep
for the string including the double quoting.
(yes, I really use grep to understand the source-code)
NM_CACHED_QUARK_FCN() is a replacement for G_DEFINE_QUARK().
G_DEFINE_QUARK() is mostly used to define GError quarks. As
such, it always appends _quark() to the function name, which
is unfavorable because it makes it harder to grep for the
definition of the function.
In general I think that macros that defined symbols by concatenating
something should be avoided because that makes it harder to locate
where the symbol was defined.
Fix it by converting the macro to an inline function. It's anyway
nicer.
$ make src/src_libNetworkManagerBase_la-main-utils.lo
CC src/src_libNetworkManagerBase_la-main-utils.lo
In file included from ./shared/nm-utils/nm-macros-internal.h:29:0,
from ./shared/nm-default.h:178,
from src/main-utils.c:22:
src/main-utils.c: In function ‘nm_main_utils_setup_signals’:
./shared/nm-utils/nm-glib.h:144:36: error: comparison of unsigned expression >= 0 is always true [-Werror=type-limits]
&& glib_micro_version >= (micro))))
^
src/main-utils.c:82:6: note: in expansion of macro ‘nm_glib_check_version’
if (nm_glib_check_version (2, 36, 0)) {
^~~~~~~~~~~~~~~~~~~~~
cc1: all warnings being treated as errors
Makefile:12312: recipe for target 'src/src_libNetworkManagerBase_la-main-utils.lo' failed
Similar to systemd's PROTECT_ERRNO. The difference it, that it doesn't
treat the auto-variable as internal, so it is allowed to use it. E.g.
if (!(fd = open (...)) {
NM_AUTO_PROTECT_ERRNO (errno_saved);
printf ("error: %s", g_strerror (errno_saved));
return FALSE;
}
We already have gs_fd_close, which however doesn't preserve
errno and only checks for fd != -1. Add our own define.
Downside is, we have to include stdio.h and errno.h,
which effectively ends up to be included *everywhere*.
We usually don't build NM with g_assert() disabled (G_DISABLE_ASSERT).
But even if we would, there is no assertion macro that always evaluates
the condition for possible side effects.
I think that is a useful thing to have.
It was a macro to pass on the non-const-ness of the argument, but
that just doesn't make sense.
That is a signature
char *nm_str_not_empty (char *)
does not make sense, because you cannot transfer ownership
conditionally without additional checks to avoid a leak. Which makes
this form is pointless. For example:
char *
foo (void)
{
char *s;
s = _create_value ();
return nm_str_not_empty (s); /* leaks "" */
}
g_assert() uses G_LIKELY(), which in turn uses _G_BOOLEAN_EXPR().
As glib's version of _G_BOOLEAN_EXPR() uses a local variable
_g_boolean_var_, we cannot nest a G_LIKELY() inside a G_LIKELY(),
or inside a g_assert(), or a g_assert() inside a g_assert().
Workaround that, by redefining the macro.
I already encountered this problem before, when having a nm_assert()
inside a ({...}) block, inside a g_assert(). Then I just avoided that
combination, but this situation is quite easy to encounter.
When a VPN plugin logs to syslog(), it should not use the syslog
levels that were passed in by NetworkManager directly. Instead,
it must map LOG_NOTICE to LOG_INFO and LOG_INFO to LOG_DEBUG.
Add a utility function does gets that right.
Soon we will add proxy support where VPN plugins set a property
NM_VPN_PLUGIN_CONFIG_PROXY_PAC.
All a VPN plugin needs to make use of this new setting is the
NM_VPN_PLUGIN_CONFIG_PROXY_PAC define.
We don't want that older plugins (still compatible with libnm 1.2 API)
require a new API only for this define. Define it instead in
"shared/nm-utils/nm-vpn-plugin-macros.h" as fallback.
https://mail.gnome.org/archives/networkmanager-list/2016-June/msg00154.html
