Re: urlmon: Compile tests with -D__WINESRC__.

[Dmitry Timoshkov]

Jacek Caban  wrote:
> -
>            expectedLen = lstrlen(prop.value);
> -
>            brokenLen = lstrlen(prop.broken_value);
> -
>            expectedLen = lstrlenA(prop.value);
>
>            /* This won't be necessary once GetPropertyLength is implemented. */
>            receivedLen = -1;
>
> @@ -8145,14 +8144,14 @@ static void test_IUri_GetPropertyLength(void) {
>                                  hr, prop.expected, i, j);
>                      }
>                      todo_wine {
> -
>                    ok(receivedLen == expectedLen || broken(receivedLen == brokenLen),
> -
>                    ok(receivedLen == expectedLen || broken(prop.broken_value && receivedLen == lstrlenA(prop.broken_value)),

Please don't use lstrlenA in the tests.

urlmon: Compile tests with -D__WINESRC__.

[Saulius Krasuckas]

* On Fri, 8 Nov 2013, Dmitry Timoshkov wrote:
> > @@ -8145,14 +8144,14 @@ static void test_IUri_GetPropertyLength(void) {
> >                                  hr, prop.expected, i, j);
> >                      }
> >                      todo_wine {
> > -
> >                    ok(receivedLen == expectedLen || broken(receivedLen == brokenLen),
> > -
> >                    ok(receivedLen == expectedLen || broken(prop.broken_value && receivedLen == lstrlenA(prop.broken_value)),
>
> Please don't use lstrlenA in the tests.

Dmitry, could you please elaborate more on that?

S.

urlmon: Compile tests with -D__WINESRC__.

[Dmitry Timoshkov]

Saulius Krasuckas  wrote:
> > Please don't use lstrlenA in the tests.
>
> Dmitry, could you please elaborate more on that?

It hides the bugs.

urlmon: Compile tests with -D__WINESRC__.

[Alexandre Julliard]

Dmitry Timoshkov  writes:
> Saulius Krasuckas  wrote:
> > > Please don't use lstrlenA in the tests.
> >
> > Dmitry, could you please elaborate more on that?
>
> It hides the bugs.

Not with -D__WINESRC__.

urlmon: Compile tests with -D__WINESRC__.

[Dmitry Timoshkov]

Alexandre Julliard  wrote:
> > > > Please don't use lstrlenA in the tests.
> > >
> > > Dmitry, could you please elaborate more on that?
> >
> > It hides the bugs.
>
> Not with -D__WINESRC__.

PSDK or mingw headers don't have an idea about __WINESRC__.

urlmon: Compile tests with -D__WINESRC__.

[Alexandre Julliard]

Dmitry Timoshkov  writes:
> Alexandre Julliard  wrote:
> > > > > Please don't use lstrlenA in the tests.
> > > >
> > > > Dmitry, could you please elaborate more on that?
> > >
> > > It hides the bugs.
> >
> > Not with -D__WINESRC__.
>
> PSDK or mingw headers don't have an idea about __WINESRC__.

Yes but it can't possibly hide bugs, since all tests will be built and
tested with the Wine headers too.

urlmon: Compile tests with -D__WINESRC__.

[Dmitry Timoshkov]

Alexandre Julliard  wrote:
> > > > > > Please don't use lstrlenA in the tests.
> > > > >
> > > > > Dmitry, could you please elaborate more on that?
> > > >
> > > > It hides the bugs.
> > >
> > > Not with -D__WINESRC__.
> >
> > PSDK or mingw headers don't have an idea about __WINESRC__.
>
> Yes but it can't possibly hide bugs, since all tests will be built and
> tested with the Wine headers too.

I don't follow, it wouldn't hide bugs only if the tests would be built and
tested *without* the Wine headers too.

urlmon: Compile tests with -D__WINESRC__.

[Alexandre Julliard]

Dmitry Timoshkov  writes:
> Alexandre Julliard  wrote:
> > > > > > > Please don't use lstrlenA in the tests.
> > > > > >
> > > > > > Dmitry, could you please elaborate more on that?
> > > > >
> > > > > It hides the bugs.
> > > >
> > > > Not with -D__WINESRC__.
> > >
> > > PSDK or mingw headers don't have an idea about __WINESRC__.
> >
> > Yes but it can't possibly hide bugs, since all tests will be built and
> > tested with the Wine headers too.
>
> I don't follow, it wouldn't hide bugs only if the tests would be built and
> tested *without* the Wine headers too.

What hides bugs is the exception handler in lstrlen. When the tests are
built with Wine headers, there's no exception handler, so it will crash
and the bug will be caught. The only way a bug could remain hidden is if
no one ever builds that test with Wine headers, which is not realistic.

urlmon: Compile tests with -D__WINESRC__.

[Dmitry Timoshkov]

Alexandre Julliard  wrote:
> > > > > > > > Please don't use lstrlenA in the tests.
> > > > > > > 
> > > > > > > Dmitry, could you please elaborate more on that?
> > > > > >
> > > > > > It hides the bugs.
> > > > >
> > > > > Not with -D__WINESRC__.
> > > >
> > > > PSDK or mingw headers don't have an idea about __WINESRC__.
> > >
> > > Yes but it can't possibly hide bugs, since all tests will be built and
> > > tested with the Wine headers too.
> >
> > I don't follow, it wouldn't hide bugs only if the tests would be built and
> > tested *without* the Wine headers too.
>
> What hides bugs is the exception handler in lstrlen. When the tests are
> built with Wine headers, there's no exception handler, so it will crash
> and the bug will be caught. The only way a bug could remain hidden is if
> no one ever builds that test with Wine headers, which is not realistic.

That's correct, but it would be much better IMO to avoid any possible
difference in the generated test binaries with different compilers and
header sets. That's why any hidden dependency on __WINESRC__ on any other
Wine specific things in headers should be avoided as much as possible.