Re: server: Try2: add case-sensitivity comparison flag to find_object & co

[Alexandre Julliard]

Vitaliy Margolen  writes:
> I guess this one got lost. Resend it.
> Or is there anything wrong with it?

Well, you didn't change create_named_object, which looks
suspicious. Also I'd really like to see some regression tests for the
behavior of the various objects.

server: Try2: add case-sensitivity comparison flag to find_object & co

[Vitaliy Margolen]

Monday, September 19, 2005, 7:47:08 AM, Alexandre Julliard wrote:
> Vitaliy Margolen  writes:
> > I guess this one got lost. Resend it.
> > Or is there anything wrong with it?
> Well, you didn't change create_named_object, which looks
> suspicious. Also I'd really like to see some regression tests for the
> behavior of the various objects.

Ah, I see. I new I was missing something. As far as regression testing - most
of the named objects created from kernel32 and they are case insensitive. I
purposely left it case sensitive to keep it same way as it was before. I know of
at least named pipes that fail with current code, if I start mixing case - which
does work on windows.

This is a first step to making it right. The next step would be to pass obj_attr
attributes to server so it can check for OBJ_CASE_INSENSITIVE.

I will run all tests to check if it broke anything. And also add the
case_sensitive flag to create_named_object as well.

Vitaliy

server: Try2: add case-sensitivity comparison flag to find_object & co

[Alexandre Julliard]

Vitaliy Margolen  writes:
> Ah, I see. I new I was missing something. As far as regression testing - most
> of the named objects created from kernel32 and they are case insensitive. I
> purposely left it case sensitive to keep it same way as it was before. I know of
> at least named pipes that fail with current code, if I start mixing case - which
> does work on windows.

That's why we need regression tests. If things currently fail you can
add todo_wine, but we need to know what the exact behavior is supposed
to be.

server: Try2: add case-sensitivity comparison flag to find_object & co

[Robert Shearman]

Vitaliy Margolen wrote:
> Monday, September 19, 2005, 7:47:08 AM, Alexandre Julliard wrote:
> > Vitaliy Margolen  writes:
> > > I guess this one got lost. Resend it.
> > > Or is there anything wrong with it?
> > Well, you didn't change create_named_object, which looks
> > suspicious. Also I'd really like to see some regression tests for the
> > behavior of the various objects.
>
> Ah, I see. I new I was missing something. As far as regression testing - most
> of the named objects created from kernel32 and they are case insensitive. I
> purposely left it case sensitive to keep it same way as it was before. I know of
> at least named pipes that fail with current code, if I start mixing case - which
> does work on windows.

Something like this should test the behaviour (written in MSVC so will 
need some cleaning up before adding to a Wine test):

WCHAR buffer1[] = L"\BaseNamedObjects\Test";
    WCHAR buffer2[] = L"\BaseNamedObjects\test";
    WCHAR buffer3[] = L"\BaseNamedObjects\TEst";
    HANDLE Event, Mutant, h;
    NTSTATUS status;
    OBJECT_ATTRIBUTES attr;

attr.Length = sizeof(attr);
    attr.RootDirectory = NULL;
    attr.ObjectName = HeapAlloc(GetProcessHeap(), 0, 
sizeof(UNICODE_STRING));
    attr.Attributes = 0;
    attr.SecurityDescriptor = NULL;
    attr.SecurityQualityOfService = NULL;
    attr.ObjectName->Buffer = buffer2;
    attr.ObjectName->Length = wcslen(buffer2) * sizeof(WCHAR);
    attr.ObjectName->MaximumLength = (wcslen(buffer2) + 1) * sizeof(WCHAR);
    status = NtCreateMutant(&Mutant, GENERIC_ALL, &attr, FALSE);
    attr.ObjectName->Buffer = buffer1;
    attr.ObjectName->Length = wcslen(buffer1) * sizeof(WCHAR);
    attr.ObjectName->MaximumLength = (wcslen(buffer1) + 1) * sizeof(WCHAR);
    status = NtCreateEvent(&Event, GENERIC_ALL, &attr, FALSE, FALSE);
    attr.Attributes = OBJ_CASE_INSENSITIVE;
    attr.ObjectName->Buffer = buffer3;
    attr.ObjectName->Length = wcslen(buffer3) * sizeof(WCHAR);
    attr.ObjectName->MaximumLength = (wcslen(buffer3) + 1) * sizeof(WCHAR);
    status = NtOpenMutant(&h, GENERIC_ALL, &attr);
    ok(status == STATUS_OBJECT_TYPE_MISMATCH, "NtOpenMutant should have 
failed with STATUS_OBJECT_TYPE_MISMATCH because of case-insensitive 
search finding an event object\n");
    return 0;