Re: mlang: Make mlang tests pass under windows

[Dmitry Timoshkov]

"Maarten Lankhorst"  wrote:
> -
>        /* Not all codepages can be converted to UTF-8 */
> -
>        if (cpinfo[i].uiCodePage == 20127 || cpinfo[i].uiCodePage == 28591 || cpinfo[i].uiCodePage == 28605)
> -
>            expected = S_FALSE;

The statement above is not true. The reason for mlang test failures is that
apparently mlang in Windows has its own idea of supported locales not connected
to real kernel32 implementation. I'd leave the tests as they are, there is
nothing wrong in them, we can't fix Windows bugs.

mlang: Make mlang tests pass under windows

[Maarten Lankhorst]

Hello Dmitry,

2008/5/7 Dmitry Timoshkov :
> "Maarten Lankhorst"  wrote:
> > -
> >        /* Not all codepages can be converted to UTF-8 */
> > -
> >        if (cpinfo[i].uiCodePage == 20127 || cpinfo[i].uiCodePage ==
>
> 28591 || cpinfo[i].uiCodePage == 28605)
> > -
> >            expected = S_FALSE;
>
> The statement above is not true. The reason for mlang test failures is that
>  apparently mlang in Windows has its own idea of supported locales not
> connected
>  to real kernel32 implementation. I'd leave the tests as they are, there is
>  nothing wrong in them, we can't fix Windows bugs.

I thought the whole points of the tests was to test what wine does compared to
what windows does, if I look at
http://test.winehq.org/data/200805062030/ it fails for pretty much all
windows versions except xp without service pack. I don't see why the
test SHOULDN'T be allowed to return S_FALSE if all recent versions of
windows do.

Cheers,
Maarten.

mlang: Make mlang tests pass under windows

[Dmitry Timoshkov]

"Maarten Lankhorst"  wrote:
> > The statement above is not true. The reason for mlang test failures is that
> >  apparently mlang in Windows has its own idea of supported locales not
> > connected
> >  to real kernel32 implementation. I'd leave the tests as they are, there is
> >  nothing wrong in them, we can't fix Windows bugs.
>
> I thought the whole points of the tests was to test what wine does compared to
> what windows does, if I look at
> http://test.winehq.org/data/200805062030/ it fails for pretty much all
> windows versions except xp without service pack. I don't see why the
> test SHOULDN'T be allowed to return S_FALSE if all recent versions of
> windows do.

Then at least try to fix this properly, not hacking the test to simply show
it has passed.