Hi Martin, on my i7 machine, your patch series fails every time for me when run normally. Here's the log:
]] ../../../tools/runtest -q -P wine -M cmd.exe -T ../../.. -p cmd.exe_test.exe.so batch.c && touch batch.ok ]] ... ]] wine: cannot find L"C:\windows\system32\tell;1;;2.exe" ]] batch.c:301: Test succeeded inside todo block: unexpected char 0x0 position -1 in line 50 (got '0 ', wanted '0@space@') ]] batch.c:301: Test failed: unexpected char 0x65 position 0 in line 53 (got 'error 9009', wanted '1@space@') ]] batch.c:301: Test failed: unexpected char 0x65 position 0 in line 59 (got 'error 9009', wanted '1@space@') ]] batch.c:301: Test failed: unexpected char 0x65 position 0 in line 62 (got 'error 9009', wanted '2@space@') ]] batch.c:301: Test succeeded inside todo block: unexpected char 0x0 position -1 in line 75 (got '"hi"', wanted '"hi"') ]] batch.c:301: Test failed: unexpected char 0x65 position 0 in line 78 (got 'error 9009', wanted '1@space@') ]] batch.c:301: Test succeeded inside todo block: unexpected char 0x0 position -1 in line 81 (got '""\"\"\\" ""\"\"\\"', wanted '""\"\"\\"@space@""\"\"\\"') ]] batch.c:301: Test succeeded inside todo block: unexpected char 0x0 position -1 in line 90 (got '1 ', wanted '1@space@') ]] batch.c:301: Test succeeded inside todo block: unexpected char 0x0 position -1 in line 93 (got '0 ', wanted '0@space@') ]] batch.c:301: Test succeeded inside todo block: unexpected char 0x0 position -1 in line 102 (got '3 ', wanted '3@space@') ]] Command exited with non-zero status 10 ]] make: *** [batch.ok] Error 10 ]] make: Target `test' not remade because of errors.
When run with WINEDEBUG=warn+heap, it passes. Not sure what to make of that, but something's fishy, and the patch should not go in as is. You might try valgrind to see if it turns anything up. - Dan
Hi Dan,
on my i7 machine, your patch series fails every time for me when run normally. Here's the log:
Did you apply only this one, or also 3/9 which is supposed to fix the test results? Frédéric already told me that I should have submitted the patches in one.
Martin
]] ../../../tools/runtest -q -P wine -M cmd.exe -T ../../.. -p cmd.exe_test.exe.so batch.c && touch batch.ok ]] ... ]] wine: cannot find L"C:\windows\system32\tell;1;;2.exe" ]] batch.c:301: Test succeeded inside todo block: unexpected char 0x0 position -1 in line 50 (got '0 ', wanted '0@space@') ]] batch.c:301: Test failed: unexpected char 0x65 position 0 in line 53 (got 'error 9009', wanted '1@space@') ]] batch.c:301: Test failed: unexpected char 0x65 position 0 in line 59 (got 'error 9009', wanted '1@space@') ]] batch.c:301: Test failed: unexpected char 0x65 position 0 in line 62 (got 'error 9009', wanted '2@space@') ]] batch.c:301: Test succeeded inside todo block: unexpected char 0x0 position -1 in line 75 (got '"hi"', wanted '"hi"') ]] batch.c:301: Test failed: unexpected char 0x65 position 0 in line 78 (got 'error 9009', wanted '1@space@') ]] batch.c:301: Test succeeded inside todo block: unexpected char 0x0 position -1 in line 81 (got '""\"\"\\" ""\"\"\\"', wanted '""\"\"\\"@space@""\"\"\\"') ]] batch.c:301: Test succeeded inside todo block: unexpected char 0x0 position -1 in line 90 (got '1 ', wanted '1@space@') ]] batch.c:301: Test succeeded inside todo block: unexpected char 0x0 position -1 in line 93 (got '0 ', wanted '0@space@') ]] batch.c:301: Test succeeded inside todo block: unexpected char 0x0 position -1 in line 102 (got '3 ', wanted '3@space@') ]] Command exited with non-zero status 10 ]] make: *** [batch.ok] Error 10 ]] make: Target `test' not remade because of errors.
When run with WINEDEBUG=warn+heap, it passes. Not sure what to make of that, but something's fishy, and the patch should not go in as is. You might try valgrind to see if it turns anything up.
- Dan
On Tue, Sep 20, 2011 at 1:17 PM, Martin Wilck mwilck@arcor.de wrote:
on my i7 machine, your patch series fails every time for me when run normally. Here's the log:
Did you apply only this one, or also 3/9 which is supposed to fix the test results? Frédéric already told me that I should have submitted the patches in one.
This happened even if I applied your whole series.
Oddly, everything was fine on my other machines, it was only the i7 that detected a problem. - Dan
On 09/20/2011 09:52 PM, Dan Kegel wrote:
This happened even if I applied your whole series.
But the log you posted can't be from the full series (tests are marked as @todo_wine@ which are solved after applying all patches).
Could you please send me the output of wine cmd /c test_cmdline.cmd, with the whole series applied?
Oddly, everything was fine on my other machines, it was only the i7 that detected a problem.
What could that mean? My stuff is pretty elementary, how could that be CPU-dependent? What does warn+heap have to do with it?
Martin
On Tue, Sep 20, 2011 at 22:17, Martin Wilck mwilck@arcor.de wrote:
Hi Dan,
on my i7 machine, your patch series fails every time for me when run normally. Here's the log:
Did you apply only this one, or also 3/9 which is supposed to fix the test results? Frédéric already told me that I should have submitted the patches in one.
Not exactly. What I said it that you can't fix stuff in patch N and alter .exp in patch N+1, since that would make the test suite run with errors Ideally, what should be done is:
Patch N .cmd add tests
.exp expected results, potentially with some @todo_wine@
Patch N+1 Fix in .c/h files, and removal of @todo_wine@ in .exp
Of course, test suite must run flawlessly after both patch N and N+1 This way you (read AJ) can be sure that there was a bug (patch N) and that patch N+1 fixed it
Only if it's too difficult/not feasible to split into 2 patches (e.g. due to big deficiencies in the test runner, etc.) should you merge them, but that's generally not recommended
Frédéric
-
Dan Kegel -
Frédéric Delanoy -
Martin Wilck