On Sat, 2014-03-22 at 12:12 -0400, David Shea wrote:
This message is only printed by the "broadway" backend, and
the broadway
backend is only used because Gdk tries every backend if it can't figure
out what is going on, so just don't use that backend. The content of the
"Unable to init server" message changed recently which is why it started
getting past our false-positive filter again.
---
tests/pylint/pylint-false-positives | 1 -
tests/pylint/runpylint.sh | 5 +++++
2 files changed, 5 insertions(+), 1 deletion(-)
diff --git a/tests/pylint/pylint-false-positives b/tests/pylint/pylint-false-positives
index e2b0c09..faa1019 100644
--- a/tests/pylint/pylint-false-positives
+++ b/tests/pylint/pylint-false-positives
@@ -83,6 +83,5 @@
^E0611:[ 0-9]*,[0-9]*: No name '_isys' in module 'pyanaconda'$
^E0611:[ 0-9]*,[0-9]*:.*: No name '_isys' in module 'pyanaconda'$
^E0712:[ 0-9]*,[0-9]*:.*: Catching an exception which doesn't inherit from
BaseException: GError$
-^Unable to init server$
gi/module.py:[0-9]*: Warning: g_hash_table_insert_internal: assertion 'hash_table !=
NULL' failed$
^ g_type = info\.get_g_type\(\)$
diff --git a/tests/pylint/runpylint.sh b/tests/pylint/runpylint.sh
index 971a8f9..0e32c3b 100755
--- a/tests/pylint/runpylint.sh
+++ b/tests/pylint/runpylint.sh
@@ -28,6 +28,11 @@ unset TERM
# Don't try to connect to the accessibility socket
export NO_AT_BRIDGE=1
+# Force the GDK backend to X11. Otherwise if no display can be found, Gdk
+# tries every backend type, which includes "broadway," which prints an error
+# and keeps changing the content of said error.
+export GDK_BACKEND=x11
This is also useful for the Anaconda itself as it removes
the error
message when we try to run GUI exception handling in text mode. The same
applies to XDG_RUNTIME_DIR. Posting a patch adding those two.
--
Vratislav Podzimek
Anaconda Rider | Red Hat, Inc. | Brno - Czech Republic