This reverts commit 72a888748e, reversing
changes made to d57ef9364a.
This merge causes really high CPU load if using both pasink and
pasource, because those two modules trigger each other, and there's not
a terrible lot I can do about that, unfortunately.
* initialize first line of output earlier (before modules are
initialized, so that module/thread output cannot interfere)
* make sure that update and draw are protected against concurrent access
attempting a different tack: Reduce the amount of draws that are being
emitted by the pulseaudio module to max. 2/s. That hopefully increases
reactivity and at the same time keeps flickering to a minimum.
see #917
according to research (Jakob Nielsen '93), roughly 0.1s is what is
required for the user to feel "instantaneous".
based on this, rate-limit updates to only once per ~0.03s (0.1 felt
really laggy for me, so let's be conservative)
Added a second check for indications of possible change of public IP address since netifaces does not properly handle all Linux routing tables (ref: http://linux-ip.net/html/routing-tables.html) which can lead to changes being missed if only looking at defauilt route. Consequently an additional check for changes to the network interface names/numbers was added to further help in identifying potentially notable changes. Note that netifaces is no longer being maintained so the underlying issue with its handling of gateways is unlikely to be resolved (https://github.com/al45tair/netifaces). An alternative would be to use direct calls to the OS like 'ip route list table all' (note the difference to 'ip route list all') but this might lead to unpredicatble results between distributions/flavours so probably best to stick with a library for now. all') but this might lead to unpredicatble results between distributions/flavours so probably best to stick with a library for now.
Introduced time.sleep(2) following calls to util.location.reset() since it can take util.location a while to update following a call to .reset() which can cause calls to .location_info() to return unpredicatable/unuseful results.
Falls in the "meant well, but doesn't really make sense" category: When
the volume exceeds 100%, the widget was shown in "critical" state. Some
headsets, audio cards, etc. do require a high volume setting, however.
And anyhow, it's really up to the user.
fixes#913
add a boolean flag ("adjust") to redshift that allows the user to have
bumblebee-status actually perform the color adjustment (by invoking
redshift in "one-shot" node).
Note that this only updates the color value each time the redshift
module is updated (every 10s, by default), and likely will collide with
any running redshift process.
fixes#908
As far as I can tell, Travis CI is *not* free anymore, even for OSS
projects. I assumed the 10k credits were monthly or yearly, but
seemingly, they are a one-time thing.
So, remove Travis CI and start searching for a better replacement.
see #906
