ole32: Accept STG_E_UNIMPLEMENTEDFUNCTION when ILockBytes implementation doesn't support locking.

[Dmitry Timoshkov]

Nikolay Sivov <bunglehead at gmail.com> wrote:

> >>> Besides, my patch fixes a regression caused by IStorage locking rewrite,
> >>> and should be pretty obvious without a test case. With this regression
> >>> fixed an application that I have here can open its database files again
> >>> as it was before.
> >>>
> >>> If you really insist on a test case, please ask Vincent, he is the original
> >>> authour of this implementation.
> >>>
> >>
> >> He's not the one sending this, so why should he be writing tests for
> >> your changes?
> >
> > My patch just adds another obvious case of a possible ILockBytes failure.
> > If the tests were not needed for an already existing case then my patch
> > changes nothing in that regard.
> 
> Asking for a test especially if you found a regression is not uncommon. 
> Your change brings inconsistency in error handling as I mentioned 
> already, and it's not obvious at all what errors should be ignored.

What inconsistency do you mean?

> I really don't understand why are you so resistant sometimes.

I just really don't undrestand why you are so selective sometimes.
Certainly I can understand your desire to have a test for something,
what I don't get is why your desire didn't appear earlier for the same
piece of code but sent by somebody else. Besides my patch is so obvious
and trivial, that requesting to write completely new set of tests is not
justified at all IMHO.

I'll probably can have stab at it when I have a bit of free time, but this
should be a separate and not related patch set.

> > Fixing a regression with a pretty obvious
> > fix shouldn't deserve so much of salt from your side Nikolay.

There is not much to add.

-- 
Dmitry.