[Bug 23095] New: The win_skip()s in msvcrt/tests/printf.c are broken on Win95/98 and NT4

http://bugs.winehq.org/show_bug.cgi?id=23095

Summary: The win_skip()s in msvcrt/tests/printf.c are broken on
                    Win95/98 and NT4
           Product: Wine
           Version: 1.1.43
          Platform: x86
               URL: http://source.winehq.org/git/wine.git/?a=blob;f=dlls/m
                    svcrt/tests/printf.c;hb=HEAD
        OS/Version: Linux
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: msvcrt
        AssignedTo: [e-mail]
        ReportedBy: [e-mail]

In commit b205fd8108eca2fd29705875e53ed852f56a6b6d, I added some win_skips to
the tests in msvcrt/tests/printf.c. The really strange thing is that on
Win95/98 and NT4, the win_skips fail, as though they are not running on
Windows:

printf.c:926: Test failed: _vscprintf not available
 printf.c:931: Test failed: _vscwprintf not available
 printf.c:936: Test failed: _vsnwprintf_s not available
 printf: 357 tests executed (0 marked as todo, 3 failures), 0 skipped.

This bug only affects printf.c. It doesn't affect any of the other files in
msvcrt/tests which use win_skip.

It didn't occur when I submitted the patch as a WineTestBot job, only when run
as part of winetest.exe.

I don't know what is causing it, so I'm filing a bug.

Random notes:

* The only machines affected all use version 6.0.8797.0 of msvcrt.dll.

* printf.c is the only msvcrt test which uses _CRT_NON_CONFORMING_SWPRINTFS
(but the tests are compiled with mingw and Wine headers (I think) so this
shouldn't have any effect).

* printf.c is the only msvcrt test which uses __ms_va_list and co. (but this
should have any affect either)

* wine-devel message:
http://www.winehq.org/pipermail/wine-devel/2010-June/084142.html

[Bug 23095] The win_skip()s in msvcrt/tests/printf.c are broken on Win95/98 and NT4

http://bugs.winehq.org/show_bug.cgi?id=23095

Alexander Scott-Johns  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |download, source
           Severity|normal                      |minor

[Bug 23095] The win_skip()s in msvcrt/tests/printf.c are broken on Win95/98 and NT4

http://bugs.winehq.org/show_bug.cgi?id=23095

Dmitry Timoshkov  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|msvcrt                      |testcases

[Bug 23095] The win_skip()s in msvcrt/tests/printf.c are broken on Win95/98 and NT4

http://bugs.winehq.org/show_bug.cgi?id=23095

Paul Vriens  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[e-mail]

--- Comment #1 from Paul Vriens   2010-06-10 11:28:06 ---
Did you try running "winetest.exe mscvrt:file" on a Windows box and see what
the results are (default output is somewhere in your %TMP% directory) ?

[Bug 23095] The win_skip()s in msvcrt/tests/printf.c are broken on Win95/98 and NT4

http://bugs.winehq.org/show_bug.cgi?id=23095

--- Comment #2 from Paul Vriens   2010-06-10 12:04:53 ---
Let me answer my own question:

C:\wct>msvcrt_test.exe printf
printf.c:844: Tests skipped: _vscprintf not available
printf.c:871: Tests skipped: _vscwprintf not available
printf.c:902: Tests skipped: _vsnwprintf_s not available
printf: 354 tests executed (0 marked as todo, 0 failures), 3 skipped.

C:\wct>d:\winetest.exe msvcrt:printf -o res
C:\wct>type res
....
....
Test output:
msvcrt:printf start dlls/msvcrt/tests/printf.c -
printf.c:844: Test failed: _vscprintf not available
printf.c:871: Test failed: _vscwprintf not available
printf.c:902: Test failed: _vsnwprintf_s not available
printf: 357 tests executed (0 marked as todo, 3 failures), 0 skipped.
msvcrt:printf done (3)

[Bug 23095] The win_skip()s in msvcrt/tests/printf.c are broken on Win95/98 and NT4

http://bugs.winehq.org/show_bug.cgi?id=23095

--- Comment #3 from Alexander Scott-Johns   2010-06-10 14:01:44 ---
(In reply to comment #1)
> Did you try running "winetest.exe mscvrt:file" on a Windows box and see what
> the results are (default output is somewhere in your %TMP% directory) ?

I have Windows 95 installed in a VM, but for some reason it doesn't recognise
mounted floppy disk images (it did before; not sure what broke). Is there
another way to get data into a VM image in VirtualBox? I don't want to
reinstall it as stock Win95 doesn't even come with msvcrt.dll.

Also, how do you build winetest.exe or msvcrt_test.exe?
In programs/winetest, "make winetest.exe" produces a shell script.
In dlls/msvcrt/tests, "make crosstest" fails to link ("cannot find
-lwinecrt0").

(In reply to comment #2)
> Let me answer my own question:
>
> C:\wct>msvcrt_test.exe printf
> printf.c:844: Tests skipped: _vscprintf not available
> printf.c:871: Tests skipped: _vscwprintf not available
> printf.c:902: Tests skipped: _vsnwprintf_s not available
> printf: 354 tests executed (0 marked as todo, 0 failures), 3 skipped.
>
> C:\wct>d:\winetest.exe msvcrt:printf -o res
> C:\wct>type res
> ....
> ....
> Test output:
> msvcrt:printf start dlls/msvcrt/tests/printf.c -
> printf.c:844: Test failed: _vscprintf not available
> printf.c:871: Test failed: _vscwprintf not available
> printf.c:902: Test failed: _vsnwprintf_s not available
> printf: 357 tests executed (0 marked as todo, 3 failures), 0 skipped.
> msvcrt:printf done (3)

Why do these behave differently?
Doesn't the latter just extract msvcrt_test.exe and run the former?

[Bug 23095] The win_skip()s in msvcrt/tests/printf.c are broken on Win95/98 and NT4

http://bugs.winehq.org/show_bug.cgi?id=23095

--- Comment #4 from Paul Vriens   2010-06-10 15:11:23 ---
(In reply to comment #3)
> Also, how do you build winetest.exe or msvcrt_test.exe?
> In programs/winetest, "make winetest.exe" produces a shell script.
> In dlls/msvcrt/tests, "make crosstest" fails to link ("cannot find
> -lwinecrt0").

I used the winetest.exe from test.winehq.org. You can extract the tests by
doing 'winetest.exe -x'.

I do however compile my own winetest.exe (see also
http://wiki.winehq.org/CompilingDLLsUsingMingw)
> Why do these behave differently?
> Doesn't the latter just extract msvcrt_test.exe and run the former?

Yes, but I guess in a different context (subprocess and such).

I did find the culprit though. If I comment out the _ecvt test loop in
test_xcvt(), I get the expected "Tests skipped" in both case (single exe and
via winetest.exe).

No idea how to fix that yet. Will try and see if I can find the offending
single test(s).

[Bug 23095] The win_skip()s in msvcrt/tests/printf.c are broken on Win95/98 and NT4

http://bugs.winehq.org/show_bug.cgi?id=23095

--- Comment #5 from Paul Vriens   2010-06-10 15:26:11 ---
Created an attachment (id=28722)
 --> (http://bugs.winehq.org/attachment.cgi?id=28722)
Diff showing the problematic tests for _evct

The attached diff shows which test messes up the printf test output on win9x
and NT4. Removing this single test indeed gives me back the expected "Tests
skipped" instead of "Tests failed".

[Bug 23095] The win_skip()s in msvcrt/tests/printf.c are broken on Win95/98 and NT4

http://bugs.winehq.org/show_bug.cgi?id=23095

--- Comment #6 from Sven   2010-07-19 19:46:33 ---
Fixed by
http://source.winehq.org/git/wine.git/?a=commit;h=913c91fd470f08342f2c07bbae...

[Bug 23095] The win_skip()s in msvcrt/tests/printf.c are broken on Win95/98 and NT4

http://bugs.winehq.org/show_bug.cgi?id=23095

Austin English  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |FIXED

--- Comment #7 from Austin English   2010-07-19 22:01:57 ---
Fixed.

[Bug 23095] The win_skip()s in msvcrt/tests/printf.c are broken on Win95/98 and NT4

http://bugs.winehq.org/show_bug.cgi?id=23095

Alexandre Julliard  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |CLOSED

--- Comment #8 from Alexandre Julliard   2010-07-30 12:58:03 ---
Closing bugs fixed in 1.3.0.