Several test scripts asume ::argv0 to always be there,
which is not the case for a slave interpreter. The
tests run fine by themselves but fail from under the
all.tcl harness... This one-line patch addresses the
problem.
-mi
P.S. Some other tests (mime decode/encode) fail because
they assume the C/iso-8859-1 locale. To pass they must
run with LANG=C set...
Logged In: YES
user_id=173641
SourceForge wouldn't let me attach the file. Here it is
(watch out for spaces vs. tabs):
--- all.tcl Mon Jan 21 12:55:22 2002
+++ all.tcl Mon Jan 28 14:30:41 2002
@@ -122,4 +122,5 @@
# import the auto_path from the parent interp, so
"package require" works
$c eval {
+ set ::argv0 [pSet ::argv0]
set ::tcllibModule [pSet module]
set auto_path [pSet auto_path]
Logged In: YES
user_id=202636
Duplicate of bug #507242 except for the locale issue.
Hmm. How do we ensure the C locale from within Tcl during a
test?
Logged In: YES
user_id=75003
True regarding duplication.
However this patch seems to be more general than the patch
in #507242. Application stops future problems with argv0
immediately.
What's your opinion ?
Logged In: YES
user_id=202636
This certainly seems like a reasonable patch. It appears
that none of the current tests depend upon argv0 now but
this would prevent future problems.
I don't know about the locale issue though. I've tried
running the mime tests with LANG set to fr, de and C but it
has passed every time.
I've committed this patch as it doesn't cause any new failures.
Need to confirm the locale issue either on a unix system or
have someone who natively uses a non-english locale to check
for any such depencency.
Logged In: YES
user_id=173641
FIY. My LANG is set to uk_UA.KOI8-U. I observed failures
on FreeBSD. The FreeBSD port now sets LANG to C prior
to running tests (or anything), which provides the
work-around.
Logged In: YES
user_id=202636
Renaming this - we now use 'sak.tcl' to run the test suite -
but I'm certain the same problems will be present. Need to
retry with
tclsh sak.tcl test
Logged In: YES
user_id=202636
Moved this from patches to bugs as the patch has been
applied and this is now a bug discussion.
Logged In: YES
user_id=75003
Has the discussion closed ? As the patch has been applied,
can the entry be closed as well ? What issues are open and
unresolved ?