Re: [2/2] kernel32: Remove a 0-length read optimization from ReadFile.

[Alexandre Julliard]

Dmitry Timoshkov  writes:
> This matches WriteFile behaviour and ensures that invalid parameters are
> properly handled and fields in the OVERLAPPED structure are initialized.

This is again causing test failures. Please do a full test run before
sending patches.

[2/2] kernel32: Remove a 0-length read optimization from ReadFile.

[Dmitry Timoshkov]

Alexandre Julliard  wrote:
> > This matches WriteFile behaviour and ensures that invalid parameters are
> > properly handled and fields in the OVERLAPPED structure are initialized.
>
> This is again causing test failures. Please do a full test run before
> sending patches.

Could you please list at least some of the observed failures? That's
just copy/paste on your side but it would really save quite a bit of
time for me since not all tests pass on my system, particularly d3d
related, crypt32 and some others. And yes, I always do 'make test' in
the dlls directories that may be affected.

[2/2] kernel32: Remove a 0-length read optimization from ReadFile.

[Alexandre Julliard]

Dmitry Timoshkov  writes:
> Alexandre Julliard  wrote:
> > > This matches WriteFile behaviour and ensures that invalid parameters are
> > > properly handled and fields in the OVERLAPPED structure are initialized.
> >
> > This is again causing test failures. Please do a full test run before
> > sending patches.
>
> Could you please list at least some of the observed failures? That's
> just copy/paste on your side but it would really save quite a bit of
> time for me since not all tests pass on my system, particularly d3d
> related, crypt32 and some others. And yes, I always do 'make test' in
> the dlls directories that may be affected.

Obviously, ntdll and kernel changes can affect everything. In this case
shlwapi was the first failing one.