Re: [(try 3) 3/3] opengl32/tests: Improve test for wglMakeCurrent.

[Paul Vriens]

On 09/19/2009 12:29 PM, Rico Schüller wrote:
> dlls/opengl32/tests/opengl.c | 10 ++++++++++
> 1 files changed, 10 insertions(+), 0 deletions(-)

+    ret = wglMakeCurrent( NULL, NULL );
+    ok( !ret, "wglMakeCurrent failed\n" );

shouldn't that message read "wglMakeCurrent succeeded" ?

If you are testing the last error it's usually a good idea to set the 
last error before the call (aka "SetLastError(0xdeadbeef)").

[(try 3) 3/3] opengl32/tests: Improve test for wglMakeCurrent.

[Rico Schüller]

Am 19.09.2009 12:46, schrieb Paul Vriens:
> On 09/19/2009 12:29 PM, Rico Schüller wrote:
> > dlls/opengl32/tests/opengl.c | 10 ++++++++++
> > 1 files changed, 10 insertions(+), 0 deletions(-)
> - ret = wglMakeCurrent( NULL, NULL );
> - ok( !ret, "wglMakeCurrent failed\n" );
>
> shouldn't that message read "wglMakeCurrent succeeded" ?

Yes, probably something like "wglMakeCurrent succeeded, but should 
fail!" is really a better solution.
> If you are testing the last error it's usually a good idea to set the 
> last error before the call (aka "SetLastError(0xdeadbeef)").

I haven't called SetLastError() before, because it's not done in the 
rest of the tests in dlls/opengl32/tests/opengl.c . But if it is usually 
a good idea, I could change that, too.

[(try 3) 3/3] opengl32/tests: Improve test for wglMakeCurrent.

[Paul Vriens]

On 09/19/2009 01:15 PM, Rico Schüller wrote:
> Am 19.09.2009 12:46, schrieb Paul Vriens:
> > On 09/19/2009 12:29 PM, Rico Schüller wrote:
> > > dlls/opengl32/tests/opengl.c | 10 ++++++++++
> > > 1 files changed, 10 insertions(+), 0 deletions(-)
> > - ret = wglMakeCurrent( NULL, NULL );
> > - ok( !ret, "wglMakeCurrent failed\n" );
> >
> > shouldn't that message read "wglMakeCurrent succeeded" ?
>
> Yes, probably something like "wglMakeCurrent succeeded, but should
> fail!" is really a better solution.
> > If you are testing the last error it's usually a good idea to set the
> > last error before the call (aka "SetLastError(0xdeadbeef)").
>
> I haven't called SetLastError() before, because it's not done in the
> rest of the tests in dlls/opengl32/tests/opengl.c . But if it is usually
> a good idea, I could change that, too.

The main reason being that you don't know if the last error was set or 
changed unless you know what is was before.

[(try 3) 3/3] opengl32/tests: Improve test for wglMakeCurrent.

[Henri Verbeet]

2009/9/19 Rico Schüller :
> Yes, probably something like "wglMakeCurrent succeeded, but should fail!" is
> really a better solution.

For what it's worth, I think the exclamation mark might be a bit
overly dramatic. It's not like e.g. your printer is on fire. :-)

[(try 3) 3/3] opengl32/tests: Improve test for wglMakeCurrent.

[Detlef Riekenberg]

On Sa, 2009-09-19 at 13:15 +0200, Rico Schüller wrote:
> > - ret = wglMakeCurrent( NULL, NULL );
> > - ok( !ret, "wglMakeCurrent failed\n" );
> >
> > shouldn't that message read "wglMakeCurrent succeeded" ?
>
> Yes, probably something like "wglMakeCurrent succeeded, but should 
> fail!" is really a better solution

Please remenber, that all failures have the prefix:
<filename>:<__LINE__>: Test failed:

My style for the ok message is to print the returned result
followed by the expected result in round brackets.