On Wed, Oct 19 2016, Ruben Astudillo wrote: > On 19/10/16 08:19, Michael Alan Dorman wrote: >> Rodney Lorrimar <d...@rodney.id.au> writes: >>> >>> DBUS_SESSION_BUS_ADDRESS=unix:abstract=/tmp/dbus-jp8hFEY1dV,guid=c6e01dc144bda2ae4ca47c84580747a5;unix:abstract=/tmp/dbus-BjlHWWC2xz,guid=3733c1a431ad4cb309cc622a580747a5 >>> >>> I'm not quite sure why the dbus socket was unavailable, but my solution >>> was to enable services.dbus.socketActivated in configuration.nix. >> >> I have seen this, too---manifesting as an taffybar not starting---though >> I was doing some more investigation before I enquired. >> >> I was able to roll back to a version that worked, and the clearly >> distinguishing feature was that there was only one listing in >> DBUS_SESSION_BUS_ADDRESS; and indeed, if I set the environment variable >> when starting taffybar, it works fine. >> >> $ >> DBUS_SESSION_BUS_ADDRESS=unix:abstract=/tmp/dbus-jp8hFEY1dV,guid=c6e01dc144bda2ae4ca47c84580747a5 >> taffybar >> >> Works just fine. >> >> I guess the question I'm left with is whether the error is in the nix >> configuration providing us with two entries in the variable, or in the >> haskell dbus package in not handling it properly? > > I've noticed this too (and fixed the same way). Before in 16.03 > DBUS_SESSION_BUS_ADDRESS only listed one abstract path, thus I it isn't > a fault on the haskell side of things.
Hi, I guess it must have been a change in either dbus-launch or the dbus daemon. The current freedesktop.org implementation handles semicolons in DBUS_SESSION_BUS_ADDRESS. It also adds the user bus socket to the list of addresses (if it exists). I have reported these two things to the haskell-dbus author. Cheers, Rodney _______________________________________________ nix-dev mailing list nix-dev@lists.science.uu.nl http://lists.science.uu.nl/mailman/listinfo/nix-dev
