Hi,
I'm wondering if my patch (62325) hasn't been reviewed yet or if you expect a strict error-free patch. Indeed, http://source.winehq.org/patches/data/62325.testfail shows a "No test summary line found" for WNT4WSSP6. If it's an issue, how to fix it ?
Thanks
On 06/11/2010 11:42 AM, GOUJON Alexandre wrote:
Hi,
I'm wondering if my patch (62325) hasn't been reviewed yet or if you expect a strict error-free patch. Indeed, http://source.winehq.org/patches/data/62325.testfail shows a "No test summary line found" for WNT4WSSP6. If it's an issue, how to fix it ?
Thanks
Hi,
Those are the results from the testbot:
http://testbot.winehq.org/JobDetails.pl?Key=2467&scrshot_202=1#k202
This means this new test is crashing on (that) NT4.
Le 11/06/2010 11:58, Paul Vriens a écrit :
Those are the results from the testbot:
I know, I received the log by mail and I submitted my patch before sending it to wine-patches but I didn't understand it as an error.
This means this new test is crashing on (that) NT4.
WNT4WSSP6.log advapi32:security start - - security.c:1427: Tests skipped: CreateWellKnownSid not available security.c:1084: AccessCheck with MAXIMUM_ALLOWED got Access 0x00060019 security.c:1140: Tests skipped: AddAccessAllowedAceEx is not available security.c:1223: Tests skipped: ConvertSidToStringSidA is not available security.c:1997: Tests skipped: NULL account name doesn't work on NT4 security.c:2073: Tests skipped: CreateWellKnownSid not available advapi32:security done (3221225477)
No test summary line found
3221225477 = 0xC0000005 = STATUS_ACCESS_VIOLATION but where does it come from ? There is neither backtrace nor timeout log/explanation.
Whatever its origin, the next test after test_LookupAccountName() is test_security_descriptor() whereas my patch just modified test_GetNamedSecurityInfoA().
And a last question : Does win_skip do return ? Some tests add a return after it, and others no.
Thanks
On 06/11/2010 01:13 PM, GOUJON Alexandre wrote:
Le 11/06/2010 11:58, Paul Vriens a écrit :
Those are the results from the testbot:
I know, I received the log by mail and I submitted my patch before sending it to wine-patches but I didn't understand it as an error.
This means this new test is crashing on (that) NT4.
WNT4WSSP6.log advapi32:security start - - security.c:1427: Tests skipped: CreateWellKnownSid not available security.c:1084: AccessCheck with MAXIMUM_ALLOWED got Access 0x00060019 security.c:1140: Tests skipped: AddAccessAllowedAceEx is not available security.c:1223: Tests skipped: ConvertSidToStringSidA is not available security.c:1997: Tests skipped: NULL account name doesn't work on NT4 security.c:2073: Tests skipped: CreateWellKnownSid not available advapi32:security done (3221225477)
No test summary line found
3221225477 = 0xC0000005 = STATUS_ACCESS_VIOLATION but where does it come from ? There is neither backtrace nor timeout log/explanation.
Submit your patch to the testbot and run them on NT4 only. Make sure you leave the WINETEST_DEBUG flag and set the WINETEST_REPORT_SUCCESS one. That should show you the last successful ok-test.
Whatever its origin, the next test after test_LookupAccountName() is test_security_descriptor() whereas my patch just modified test_GetNamedSecurityInfoA().
And a last question : Does win_skip do return ?
win_skip is just to show (on test.winehq.org) that one or more tests are skipped. The difference between skip and win_skip is that skip can be used on all platforms, win_skip only on Windows (it will show a failure on Wine).
Some tests add a return after it, and others no.
That depends on the kinds of tests. If you want to test a function and the OS doesn't implement it than it makes no sense to continue so you return. If you only need to skip a particular test you do continue.
On 06/11/2010 01:20 PM, Paul Vriens wrote:
On 06/11/2010 01:13 PM, GOUJON Alexandre wrote:
Le 11/06/2010 11:58, Paul Vriens a écrit :
Those are the results from the testbot:
I know, I received the log by mail and I submitted my patch before sending it to wine-patches but I didn't understand it as an error.
This means this new test is crashing on (that) NT4.
WNT4WSSP6.log advapi32:security start - - security.c:1427: Tests skipped: CreateWellKnownSid not available security.c:1084: AccessCheck with MAXIMUM_ALLOWED got Access 0x00060019 security.c:1140: Tests skipped: AddAccessAllowedAceEx is not available security.c:1223: Tests skipped: ConvertSidToStringSidA is not available security.c:1997: Tests skipped: NULL account name doesn't work on NT4 security.c:2073: Tests skipped: CreateWellKnownSid not available advapi32:security done (3221225477)
No test summary line found
3221225477 = 0xC0000005 = STATUS_ACCESS_VIOLATION but where does it come from ? There is neither backtrace nor timeout log/explanation.
Submit your patch to the testbot and run them on NT4 only. Make sure you leave the WINETEST_DEBUG flag and set the WINETEST_REPORT_SUCCESS one. That should show you the last successful ok-test.
Whatever its origin, the next test after test_LookupAccountName() is test_security_descriptor() whereas my patch just modified test_GetNamedSecurityInfoA().
And a last question : Does win_skip do return ?
win_skip is just to show (on test.winehq.org) that one or more tests are skipped. The difference between skip and win_skip is that skip can be used on all platforms, win_skip only on Windows (it will show a failure on Wine).
Some tests add a return after it, and others no.
That depends on the kinds of tests. If you want to test a function and the OS doesn't implement it than it makes no sense to continue so you return. If you only need to skip a particular test you do continue.
I don't know if you have an account on the testbot so I tested it for you:
https://testbot.winehq.org/JobDetails.pl?Key=2586&log_201=1&scrshot_...
What you can see is the last successful test was on line 2671. This means that most likely the first of your added tests already crashes on NT4:
2671 ok(group != NULL, "group should not be NULL\n"); 2672 2673 LocalFree(pSecDesc); 2674 2675 2676 /* NULL descriptor tests */ 2677 2678 error = pGetNamedSecurityInfoA(windows_dir, SE_FILE_OBJECT,DACL_SECURIT 2679 NULL, NULL, NULL, NULL, NULL); 2680 ok(error==ERROR_INVALID_PARAMETER, "GetNamedSecurityInfo failed with er 2681
Le 11/06/2010 13:54, Paul Vriens a écrit :
I don't know if you have an account on the testbot so I tested it for you:
"I submitted [to wineTestBot] my patch before sending it to wine-patches" But many thanks for your help !!!
In that case, we have to use broken(), haven't we ?
On 06/11/2010 02:19 PM, GOUJON Alexandre wrote:
Le 11/06/2010 13:54, Paul Vriens a écrit :
I don't know if you have an account on the testbot so I tested it for you:
"I submitted [to wineTestBot] my patch before sending it to wine-patches" But many thanks for your help !!!
In that case, we have to use broken(), haven't we ?
broken() can only be used to cater for a wrong return value (or parameter in general). It won't protect against a crash as the test will be run regardless.
You need to find a way to detect NT4 (but not by checking the version) and skip those tests that crash.
-
GOUJON Alexandre -
Paul Vriens