Re: comctl32/ipaddress: Use ok() test instead of assert()

[Paul Vriens]

Paul Vriens wrote:
> Nikolay Sivov wrote:
> > This test fails to create window sometimes on Win95,
> > let's turn it into general failure.
> >
> > Changelog:
> >     - replace assert() with ok() test
> > > From b63fc6defb497505ecaab4921449327e875ab252 Mon Sep 17 00:00:00 2001
> >
> > From: Nikolay Sivov 
> > Date: Wed, 27 May 2009 14:26:37 +0400
> > Subject: Use ok() test instead of assert()
> >
> > dlls/comctl32/tests/ipaddress.c |    3 +--
> >  1 files changed, 1 insertions(+), 2 deletions(-)
> >
> > diff --git a/dlls/comctl32/tests/ipaddress.c 
> > b/dlls/comctl32/tests/ipaddress.c
> > index d04223c..b8f30e4 100644
> > --- a/dlls/comctl32/tests/ipaddress.c
> > +++ b/dlls/comctl32/tests/ipaddress.c
> > @@ -33,8 +33,6 @@ static HWND create_ipaddress_control (void)
> >      handle = CreateWindowEx(0, WC_IPADDRESS, NULL,
> >                  WS_BORDER|WS_VISIBLE, 0, 0, 0, 0,
> >                  NULL, NULL, NULL, NULL);
> > - assert(handle);
> > - return handle;
> >
> > }
> >
> > @@ -45,6 +43,7 @@ static void test_get_set_text(void)
> >      INT r;
> >  hwnd = create_ipaddress_control();
> > - ok(hwnd != NULL, "Expected window to be created\n");
>
> Shouldn't you insert a return here when hwnd is NULL. The following 
> tests will fail as well I guess.

Just had a look and it seems that only one box actually has this 
failure. Maybe a win_skip() would be 'better'?

hwnd = create_ipaddress_control();
if (!hwnd)
{
     win_skip();
     return;
}

Maybe even printing the last error in that win_skip() ?

Especially as MSDN states:

The IP address control is implemented in version 4.71 and later of 
Comctl32.dll.

And this box has 4.70.0.1146.