summaryrefslogtreecommitdiffstats
path: root/tests/iogt.test
diff options
context:
space:
mode:
authordgp <dgp@users.sourceforge.net>2014-05-09 16:38:23 (GMT)
committerdgp <dgp@users.sourceforge.net>2014-05-09 16:38:23 (GMT)
commit8b4cb6f5d95d707abacdcfbcb75153cd12edb59f (patch)
tree58a98d03dfe74691ae98039e70c16d20d55cd092 /tests/iogt.test
parentd7461d07ad79630ccb7605ea15e47e50aaea44a9 (diff)
downloadtcl-8b4cb6f5d95d707abacdcfbcb75153cd12edb59f.zip
tcl-8b4cb6f5d95d707abacdcfbcb75153cd12edb59f.tar.gz
tcl-8b4cb6f5d95d707abacdcfbcb75153cd12edb59f.tar.bz2
Added comment explaining the "knownBug" in iogt-6.1
Diffstat (limited to 'tests/iogt.test')
-rw-r--r--tests/iogt.test9
1 files changed, 9 insertions, 0 deletions
diff --git a/tests/iogt.test b/tests/iogt.test
index d4291b3..20d7538 100644
--- a/tests/iogt.test
+++ b/tests/iogt.test
@@ -968,6 +968,15 @@ test iogt-6.0 {Push back} testchannel {
} {xxx}
test iogt-6.1 {Push back and up} {testchannel knownBug} {
+
+ # This test demonstrates the bug/misfeature in the stacked
+ # channel implementation that data can be discarded if it is
+ # read into the buffers of one channel in the stack, and then
+ # that channel is popped before anything above it reads.
+ #
+ # This bug can be worked around by always setting -buffersize
+ # to 1, but who wants to do that?
+
set f [open $path(dummy) r]
# contents of dummy = "abcdefghi..."