Detlef Riekenberg wrote:
http://test.winehq.org/data/f212579ae9a1b770ebd34cec20f95e1977bb57f0/xp_dr-a...
While reading the code for a test failure on all of my Windows installations, the current code was already strange:
5436 ok(hres == VARCMP_EQ || 5437 broken(hres == VARCMP_GT), "Expected VARCMP_EQ, got %08x\n", hres);
I feel bad for adding a "broken(hres == VARCMP_LT)".
Extending vartest and inserting a trace made the localization error obvisious: VARCMP_LT was returned when comparing "12Falsch" and "12False".
The failure for the similar test (VT_BOOL with True: "12Wahr" > "12True") was hidden by the wrong "broken(hres == VARCMP_GT)".
We have also another XP Machine that produced this failure. (xp-home-SP2-czech)
Hi Detlef,
The lstrcpyW and lstrcatW additions to init() in the tests don't work on Win95. I guess that's where the new failures come from:
http://test.winehq.org/data/ed74088d7ddf0c389e7596237add7cb751b17a0a/95_W95/...
Could you have a look?
On Mi, 2009-04-15 at 13:37 +0200, Paul Vriens wrote:
The lstrcpyW and lstrcatW additions to init() in the tests don't work on Win95. I guess that's where the new failures come from:
I removed a "broken(hres == VARCMP_GT)", and that's what VarCmp on your w95 machine returned. My second Patch for oleaut32/tests/vartest.c was not applied, otherwise we would have seen the reason in the log.
Can you please use the attached diff and post the result? Thanks
Detlef Riekenberg wrote:
On Mi, 2009-04-15 at 13:37 +0200, Paul Vriens wrote:
The lstrcpyW and lstrcatW additions to init() in the tests don't work on Win95. I guess that's where the new failures come from:
I removed a "broken(hres == VARCMP_GT)", and that's what VarCmp on your w95 machine returned. My second Patch for oleaut32/tests/vartest.c was not applied, otherwise we would have seen the reason in the log.
Can you please use the attached diff and post the result? Thanks
Here it is:
vartest.c:127: Tests skipped: No support for I8 and UI8 data types vartest.c:5452: got 0x0 and VT_BSTR(12True) for VT_INT(12), VT_BOOL(1) vartest.c:5455: got 0x2 for VT_BSTR(12True), VT_BSTR() vartest.c:5456: Test failed: Expected VARCMP_EQ, got 00000002 for VT_BSTR(12True), VT_BSTR() vartest.c:5471: got 0x0 and VT_BSTR(12False) for VT_INT(12), VT_BOOL(0) vartest.c:5474: got 0x2 for VT_BSTR(12False), VT_BSTR() vartest.c:5475: Test failed: Expected VARCMP_EQ, got 00000002 for VT_BSTR(12False), VT_BSTR() vartest: 1025087 tests executed (0 marked as todo, 2 failures), 1 skipped.
As said. I think the reason for sz12_true being empty is that you are using W-functions in the init() function.
Paul Vriens wrote:
Detlef Riekenberg wrote:
On Mi, 2009-04-15 at 13:37 +0200, Paul Vriens wrote:
The lstrcpyW and lstrcatW additions to init() in the tests don't work on Win95. I guess that's where the new failures come from:
I removed a "broken(hres == VARCMP_GT)", and that's what VarCmp on your w95 machine returned. My second Patch for oleaut32/tests/vartest.c was not applied, otherwise we would have seen the reason in the log.
Can you please use the attached diff and post the result? Thanks
Here it is:
vartest.c:127: Tests skipped: No support for I8 and UI8 data types vartest.c:5452: got 0x0 and VT_BSTR(12True) for VT_INT(12), VT_BOOL(1) vartest.c:5455: got 0x2 for VT_BSTR(12True), VT_BSTR() vartest.c:5456: Test failed: Expected VARCMP_EQ, got 00000002 for VT_BSTR(12True), VT_BSTR() vartest.c:5471: got 0x0 and VT_BSTR(12False) for VT_INT(12), VT_BOOL(0) vartest.c:5474: got 0x2 for VT_BSTR(12False), VT_BSTR() vartest.c:5475: Test failed: Expected VARCMP_EQ, got 00000002 for VT_BSTR(12False), VT_BSTR() vartest: 1025087 tests executed (0 marked as todo, 2 failures), 1 skipped.
As said. I think the reason for sz12_true being empty is that you are using W-functions in the init() function.
The attached patch fixes 'my' issue on Win95 so it's indeed the use of W-functions.
-
Detlef Riekenberg -
Paul Vriens