This avoids ellipsizing various reasonably-named services on my laptop.

I need to find a better way of making the bounds of a Shape ask Pango
how big the text is, without making bounds impure...

The first time a bus connection is shown in the diagram, its names are
drawn, ideally aligned with the previous horizontal event divider.
However, previously the first of n bus names was aligned with the
divider, so the other ones spilled over into the diagram.

This has the interesting side effect that, if the first connection to be
mentioned has more than two names, only the last two are shown because
the rest are off the top of the diagram!

I think 'headers' could be written as a self-referential list, where the
y-coordinate of each element is a function of itself, because
(headerHeight . strs) is not strict in the 'shapey' field of Header. But
that would be silly.

The parser previously accepted a well-formed well-known name, or the
string "<none>". It's more useful to just accept anything, to let us
shoe-horn other stuff into the log format as needed (until there's a
better format); in particular, this will allow the field to be reused
for the process name in the short-term.

(Taken from a patch to dbus upstream; thanks!)

The current SIGINT handling of dbus-monitor ain't making too many people
happy since it defers the exit to the next msg received -- which might
be quite some time away often enough.

This patch replaces the SIGINT handling by simply enabling line-buffered
IO for STDOUT so that even if you redirect dbus-monitor into a file no
lines get accidently lost and the effect of C-c is still immediate.

halfline came up with the great idea to use setvbuf here instead of
fflush()ing after each printf().

(Oh and the old signal handler was broken anyway, the flag should have
been of type sigatomic_t and be marked volatile)

Signed-off-by: Will Thompson <will.thompson@collabora.co.uk>

This worked for Alban, and makes sense.

This works on Josh's Fedora laptop, so it's good enough for me.

When the monitor gets forcibly disconnected from the bus, the message it
receives seems to have no sender (maybe libdbus is synthesizing it).
Let's parse it, and drop it when we're filtering out messages from the
bus.

I should have thought to create a data type for the possible changes
sooner.

run-uninstalled relies on to be called from the topdir of the project.
This patch detects the path run-uninstalled is called from, sets
bustle_datadir and calls bustle.

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>

bustle-dbus-monitor calls "dbus_connection_read_write_dispatch"
blocking and "dbus_connection_read_write_dispatch" isn't interrupted by a
signals. This means bustle-dbus-monitor needs to receive at least one
message after receiving a signal to exit.

This patch adds a timeout of 1 second to the
"dbus_connection_read_write_dispatch" call, to work around this problem.

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Will Thompson <will.thompson@collabora.co.uk>

Setup.lhs is not executable. Following the README is not possible.
"./Setup.lhs configure --user"

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Will Thompson <will.thompson@collabora.co.uk>

I broke this in fd5b7c57

The bus doesn't let you send messages to non-existant names, so
bustle-dbus-monitor will never see them. So if they're in the log, the
log's corrupt.

This makes me care less about fixing this. :-)

This doesn't catch logs where unknown names are released, for instance,
but cuts out a lot of failure cases.