This adds a small function which behaves exactly like libc's system(), but also
clears the masked signal set in the child process.
This is needed because libev 3.8 masks signals. :(
Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Julien Danjou <julien@danjou.info>
awful.widget.graph and progressbar both create an imagebox. The default is
resize = true which means the image is scaled up inside the wibox if
:set_height() is used.
Setting widget.resize = false avoids this.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Julien Danjou <julien@danjou.info>
set_value() didn't call update() and thus had no visible effect on the
progressbar.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Julien Danjou <julien@danjou.info>
This makes image.argb32() fail in a better way if it is called with a zero size.
Bug found by anrxc and reported a FS#597 and FS#599.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Julien Danjou <julien@danjou.info>
This code accesses these functions via _M['set_height'], but if these are local
functions they aren't available this way. Un-local-ing helps here.
This was noticed by anrxc when he set a widget's height to 0.9. These functions
should have caught this invalid height and ignore it, but because they weren't
actually used this error checking was bypassed.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Julien Danjou <julien@danjou.info>
correct problem matching expression with - or special regexp character
Signed-off-by: Cedric GESTES <ctaf42@gmail.com>
Signed-off-by: Julien Danjou <julien@danjou.info>
When awful.client.property.set() was called it always emitted property::floating
on the client. Instead it should emit propert::<name of the property> of course!
Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Julien Danjou <julien@danjou.info>
This adds a callback function which glib calls after it fork()'d and did all the
necessary setup. This callback function clears our signal mask.
This is necessary because libev 3.8 and later use signalfd and therefor have to
add those signals to the signal mask. Processes started through awesome would
inherit this signal mask and I can tell you, some app which ignores ctrl-c
confuses people a lot.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Julien Danjou <julien@danjou.info>
This adds a new function spawn_proc_helper() which just contains a copy of
g_spawn_command_line_async()'s source code. This means that there should be no
behavior change at all here.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Julien Danjou <julien@danjou.info>
Without this patch c.class shows the instance and vice versa.
Signed-off-by: koniu <gkusnierz@gmail.com>
Signed-off-by: Julien Danjou <julien@danjou.info>
Instead of using the given atom name it always requested the WM_NAME property of
a client.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Julien Danjou <julien@danjou.info>
Julien Danjou
Uli Schlachter
Cedric GESTES
Gregor Best
Brian Gianforcaro
koniu