Created attachment 50544 [details] [review]
Don't assume that X11 autolaunch will work if DISPLAY is unset

In practice, it won't; other forms of autolaunch (like Mac OS launchd)
might, but we can't really assert either way.

I actually wonder whether to just disable test-autolaunch altogether, unless we can find something like Debian's xvfb-run (which runs a specified app under a temporary Xvfb server). It seems to me as though this test doesn't necessarily really test anything, and at worst it spuriously fails (in a minimal environment, with inaccessible X, etc.). It'd also be easy for it to leak a dbus-daemon/dbus-launch pair, which isn't ideal.

(In reply to comment #0)
> When performing X11 autolaunch, libdbus refuses to do anything if DISPLAY
> isn't set. However, the regression test assumes that, if enabled, autolaunch
> will work.
> 
> For instance, this is a problem when building dbus in a minimal chroot with
> no X server, but with X11 support enabled (as is done in Debian).

I'd like the Debian package build to be able to run the tests, one day...

Comment on attachment 50544 [details] [review]
Don't assume that X11 autolaunch will work if DISPLAY is unset

Review of attachment 50544 [details] [review]:
-----------------------------------------------------------------

Looks good.

(In reply to comment #4)
> Comment on attachment 50544 [details] [review] [review]
> Don't assume that X11 autolaunch will work if DISPLAY is unset
> 
> Review of attachment 50544 [details] [review] [review]:
> -----------------------------------------------------------------
> 
> Looks good.

Ops, need include "dbus/dbus-sysdeps.h" to fix build, introduced by calling _dbus_getenv().

(In reply to comment #5)
> Ops, need include "dbus/dbus-sysdeps.h" to fix build, introduced by calling
> _dbus_getenv().

When I wrote the patch, this file included dbus/dbus-sysdeps.h for _dbus_setenv, but I removed that in commit 412538b3b9. Reinstated and applied, 1.7.6.