On Sunday January 29 2012 2:35:34 PM you wrote: > With Tk installed, but no X11 environment running, I get this: > > couldn't connect to display "/tmp/launch-YF2sgP/org.x:0" at > /Users/sprout/.cpan/build/Tk-804.030-w5ZSVH/blib/lib/Tk/MainWindow.pm > line 53. > MainWindow->new() at t/Tk.t line 27 > # Looks like you planned 3 tests but ran 2. > # Looks like your test exited with 2 just after 2. > > In case you havenât guessed, Iâm using Mac OS X. Can you detect that I couldn't guess that, I'm not rich enough to have a Mac. :-P :-D > error and skip the test? Iâve noticed that Tk passes its own tests with > no X11 running. I havenât yet looked to see what it does. When I unset my DISPLAY and rerun the TRL tests, I get about the same problem as you. However, when I do the same thing with Tk, I get a bunch of ugly output about it :-S Anyway, I think I've about patched this up - I'm almost afraid to make my next comment for fear of another horde forming, but my system upgraded AE in the meantime, and since AE 6.11, there has been a "die" at the top of AnyEvent::Impl::Tk, not sure if it was on purpose or not (it didn't show up in the Changelog, which makes me think not - Marc doesn't seem to be shy about what he does on purpose), and that made my tests fail even with DISPLAY set :-) > Apart from that, your patch looks fine to me. Here's the updated patch with eval's around the mainwindow functions to see if it loads or not. It does mean that the AETk patch has to explicitly try it, whereas that shouldn't normally be necessary, but them's the breaks I guess.