Only on Solaris system do I see this.
And only when running the whole tcllib
test suite. Testing the pop3 module alone
doesn't cause the failure:
==== pop3-6.1 uidl FAILED
==== Contents of test case:
dialog::dialog_set uidlFailed
set psock [pop3::open localhost ak smash
[dialog::listener]]
catch {pop3::uidl $psock} msg
pop3::close $psock
dialog::waitdone
set msg
---- Result was:
POP3 UIDL ERROR:
---- Result should have been (exact matching):
POP3 UIDL ERROR: unknown command 'UIDL'
==== pop3-6.1 FAILED
Logged In: YES
user_id=75003
I made a fix to the dialog module and also to the pop3
testsuite. This is in the head. Can you check if your
problem still happens ?
Note: The fixes I made were because the tests pop3-2.8 and
pop3-5.0 were failing consistently on Solaris. Not on Linux.
I never saw pop3-6.1 failing on solaris. And my failures
were not dependent on running the whole tcllib testsuite,
pop3 alone was sufficient. So, there are quite few
differences. I am however not excluding that it is the same
problem, just appearing in different places due to different
system speed.
In previous incarnations of the testsuite driver it was
possible that a testsuite coming before pop3 forgot to kill
a timer and it fired during the later test, smashing
something. However in the current setup each module is
tested using a new sub-process. So it should not be possible
to leak stuff between modules. The only thing I can think
off is timing differences between pop3 alone vs the whole
testsuite.
Logged In: YES
user_id=80530
I can't repro the failure
today, so either your changes
fixed it, or it's not a real
reproducible bug in the first
place.
Logged In: YES
user_id=75003
Hm. Making it low-priority then. If a repro comes good,
otherwise closing in a month or so.