re: Stop shlfileop tests crashing under win98

[Dan Kegel]

Alistair wrote:
> - /* try an invalid list, only one null terminator */
> - init_shfo_tests();
> - shfo.pFrom = "";
> - shfo.wFunc = FO_DELETE;
> - ret = SHFileOperation(&shfo);
> - ok(ret == ERROR_ACCESS_DENIED, "Expected ERROR_ACCESS_DENIED, got %d\n", ret);
> - ok(file_exists("test1.txt"), "Expected test1.txt to exist\n");
> - /* This crashes under win98 */
> - if(0)
> - {
> -
>    /* try an invalid list, only one null terminator */

Hey Allistair,
that paragraph looks like it was buggy - it should have had
test1.txt followed by a single nul and then twice MAXPATH
nonnulls and then a double nul, like the similar tests I fixed in
http://www.winehq.org/pipermail/wine-patches/2008-May/055371.html
As it is, there's no way FO_DELETE could see test1.txt to maybe
delete it.

James, do you agree?

Perhaps you could leave it enabled if you fixed it like the
ones I fixed.  (Or tried to fix; sadly, Alexandre hasn't
committed it yet. Why?)
- Dan

Stop shlfileop tests crashing under win98

[Alexandre Julliard]

"Dan Kegel"  writes:
> Perhaps you could leave it enabled if you fixed it like the
> ones I fixed.  (Or tried to fix; sadly, Alexandre hasn't
> committed it yet. Why?)

Because your patch was also making unrelated and incorrect changes to
the code. If you resubmit just the test changes they'd probably go in.

Stop shlfileop tests crashing under win98

[James Hawkins]

On Tue, Jun 10, 2008 at 7:13 AM, Dan Kegel  wrote:
> Alistair wrote:
> > - /* try an invalid list, only one null terminator */
> > - init_shfo_tests();
> > - shfo.pFrom = "";
> > - shfo.wFunc = FO_DELETE;
> > - ret = SHFileOperation(&shfo);
> > - ok(ret == ERROR_ACCESS_DENIED, "Expected ERROR_ACCESS_DENIED, got %d\n", ret);
> > - ok(file_exists("test1.txt"), "Expected test1.txt to exist\n");
> > - /* This crashes under win98 */
> > - if(0)
> > - {
> > -
> >    /* try an invalid list, only one null terminator */
>
> Hey Allistair,
> that paragraph looks like it was buggy - it should have had
> test1.txt followed by a single nul and then twice MAXPATH
> nonnulls and then a double nul, like the similar tests I fixed in
> http://www.winehq.org/pipermail/wine-patches/2008-May/055371.html
> As it is, there's no way FO_DELETE could see test1.txt to maybe
> delete it.
>
> James, do you agree?

No, the test is exactly what the comment says: invalid list (empty)
with one NULL terminator.  I'm fine with it being if-def'ed out if it
crashes in win98.

Stop shlfileop tests crashing under win98

[Dan Kegel]

On Tue, Jun 10, 2008 at 5:50 AM, James Hawkins  wrote:
> No, the test is exactly what the comment says: invalid list (empty)
> with one NULL terminator.  I'm fine with it being if-def'ed out if it
> crashes in win98.

I guess I was fooled by the line
 ok(file_exists("test1.txt"), "Expected test1.txt to exist\n");
Since you didn't feed test1.txt to FO_DELETE, you shouldn't
be testing whether it's present.