Regression Test - kernel atom test

[Dmitry Timoshkov]

"Steven Edwards" <Steven_Ed4153 at yahoo.com> wrote:

> Here is another one on my win2k sp3 box.

Same here.

> All tests compiled with cygwin

Atom test was compiled with ms vc.

> atom.c:81: Test failed: succeeded adding ff74
> atom.c:84: Test failed: succeeded adding ff74
> atom.c:81: Test failed: succeeded adding ff75
> atom.c:84: Test failed: succeeded adding ff75
> atom.c:81: Test failed: succeeded adding ff76
> atom.c:84: Test failed: succeeded adding ff76
> atom.c:81: Test failed: succeeded adding ff77

[skipped]

> atom.c:81: Test failed: succeeded adding ffff
> atom.c:84: Test failed: succeeded adding ffff

I have no failures above.

> atom.c:110: Test failed: bad buffer contents
> atom.c:112: Test failed: GlobalGetAtomNameW succeeded on short buffer
> atom.c:116: Test failed: succeded
> atom.c:118: Test failed: succeded

Same failures here.

> atom.c:120: Test failed: succeded
  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
I have no this one.

> atom.c:122: Test failed: succeded
> atom.c:140: Test failed: bad length 0
> atom.c:143: Test failed: bad buffer contents aaaaaaaaaaha?xA

Same failures here.

One more problem I have is an infinite loop in test_delete_atom()
function on the line while (!GlobalDeleteAtom( atom ));

This is correct. According to Platform SDK docs GlobalDeleteAtom()
always returns (ATOM)0. But documented way of detecting failure of
the GlobalDeleteAtom() call doesn't work for me on w2k sp3 either:

"To determine whether the function has failed, call SetLastError(ERROR_SUCCESS)
before calling GlobalDeleteAtom, then call GetLastError. If the last error code
is still ERROR_SUCCESS, GlobalDeleteAtom has succeeded."

The question is: do we have to really worry about this? Is it Microsoft,
who have broken atom APIs, and not we? Can anybody run atom tests on some
other Windows platform other than Windows2000 SP3 (SP0, SP1, SP2 are fine)
and report here the result?

-- 
Dmitry.