Community
Participate
Working Groups
+++ This bug was initially created as a clone of Bug #314849 +++ CQ:WIND00108178 CQ:WIND00380449 Build ID: Eclipse Platform 4.2.1 (Juno SR1) When launching Eclipse on Linux without DISPLAY set, I'm getting this: > unsetenv DISPLAY > ./eclipse Eclipse: An error has occurred. See the log file /folk/mober/Ecltest/4.2.1_32/eclipse/configuration/1350051850732.log. With the logfile contents saying "gtk: no more handles" among some other obscure contents. So the change from bug 314849 has slightly improved things in Eclipse 3.8 but doesn't quite help enough for an end user yet. I'm setting severity MAJOR since it happens relatively easily in a networked multi-host environment to forget setting the display, and as an end user I'm pretty lost in this case. Expected behavior: Print message to stderr, similar to xterm: xterm Xt error: Can't open display: xterm: DISPLAY is not set Note that another special case is when the DISPLAY is set but the current user has no permission using it. This happens when using an "ssh -X" session and then becoming root. Here is what xterm does in this case: > ssh -X szg-tschwarz-lx1 > sudo su - > xterm Warning: This program is an suid-root program or is being run by the root user. The full text of the error or warning message cannot be safely formatted in this environment. You may get a more descriptive message by running the program as a non-root user or by removing the suid bit on the executable. xterm Xt error: Can't open display: %s I'm hoping that improving this is simple enough to do in SR2.
CQ:WIND00381930
We now use gtk_init_with_args() to initialize GTK and output the error message if any happens. Fixed in master (4.3): http://git.eclipse.org/c/equinox/rt.equinox.framework.git/commit/?id=7d5f16454127161499a284d38d4f95f0f3b49fdd http://git.eclipse.org/c/equinox/rt.equinox.framework.git/commit/?id=0f40a9c9a066f78320b6db0a64cadcf2dbf73034 http://git.eclipse.org/c/equinox/rt.equinox.framework.git/commit/?id=7cd1a2e9cd12095eb77738a8bc282fc039a8801c http://git.eclipse.org/c/equinox/rt.equinox.binaries.git/commit/?id=d46a9b5c6d8501e84c29b4c587479772960d4109
(In reply to comment #0) > I'm hoping that improving this is simple enough to do in SR2. Silenio, do you have thoughts here. Is this something we should consider for Juno SR2?
We changed the GTK initialization call. We need to test this on all GTK platforms (AIX, Solaris, HPUX, etc) before we can consider for 4.2 SR2. Even then it seems a bit too risky just for better logging.
I'll check how the new code feels as soon as I can get hold of a Kepler M3 integration build ... should be able to test Linux and Solaris. Either way, note that this is not just about better logging. It's about either helping the user in an error condition that's not too uncommon, or leaving them clueless...