Re: shdocvw: [Patch 1/2] Implement URLSubRegQueryA [Resend]

[Alexandre Julliard]

Detlef Riekenberg  writes:
> - len = sizeof(buffer);
> - res = SHRegGetUSValueA(regpath, name, NULL, buffer,  &len, FALSE, NULL, 0);
> - if (!res) {
> -
>    lstrcpynA(out, buffer, outlen);
> -
>    if (outlen > 0) {
> -
>        out[min(outlen-1,len)] = '\0';
> -
>    }

This doesn't look right. Why do you even need a separate buffer?

shdocvw: [Patch 1/2] Implement URLSubRegQueryA [Resend]

[Detlef Riekenberg]

On Mi, 2008-10-15 at 16:53 +0200, Alexandre Julliard wrote:
> Detlef Riekenberg  writes:
> > - len = sizeof(buffer);
> > - res = SHRegGetUSValueA(regpath, name, NULL, buffer,  &len, FALSE, NULL, 0);
> > - if (!res) {
> > -
> >    lstrcpynA(out, buffer, outlen);
> > -
> >    if (outlen > 0) {
> > -
> >        out[min(outlen-1,len)] = '\0';
> > -
> >    }
>
> This doesn't look right. Why do you even need a separate buffer?

The test in Patch 2 show, that URLSubRegQueryA still succeed,
when the caller buffer is to small (the result is truncated).

I used the buffer from the caller first, but SHRegGetUSValueA fail, when
the
buffer is to small.

With the seperate buffer, SHRegGetUSValueA (and URLSubRegQueryA) only
fails,
when the value does not exist in the registry.

Dropping the seperate buffer is possible, but then the implemenation
works only, when the the caller buffer is always large enough and we
need to add a buch of todo_wine.

I used a new file for the test, because URLSubRegQueryA is not related
to an object, and ParseURLFromOutsideSource is in the pipeline.