shell32/tests: Fix Program Manager DDE Conformance Test Failures

Mikey Alexander firesword14 at yahoo.com
Sun Nov 15 09:32:56 CST 2009


You really don't want to skip this one as the api is supposed to create the window, but you can't be absolutely sure it is a failure, hence the error message being what it is.  If you are going to skip the timeout if it fails, why bother checking for the window creation at all?

The real question is what is the real worst case expected time to wait before failing because the window was not created?  I upped it to 45 seconds. 

I could have implemented a window message thread to detect the window creation, but it doesn't change the problem.  Window creation is not given a deadline.  10 seconds was obviously too short.  Several machines failed on the first longer load with 10 seconds and a couple failed most of the timeouts with 10 seconds.

I'd be more than willing to up the timeout longer, but if you really do encounter a failure, it makes the failed test take that much longer.

- Mikey

--- On Sat, 11/14/09, Nicolas Le Cam <niko.lecam at gmail.com> wrote:

> From: Nicolas Le Cam <niko.lecam at gmail.com>
> Subject: Re: shell32/tests: Fix Program Manager DDE Conformance Test Failures
> To: wine-devel at winehq.org
> Cc: firesword14 at yahoo.com
> Date: Saturday, November 14, 2009, 9:09 AM
> 2009/11/14 Mikey Alexander <firesword14 at yahoo.com>:
> > Fixes a memory leak, non connection to the progman dde
> on early windows, and timeout problems with the conformance
> tests.
> >
> > ---
> >  dlls/shell32/tests/progman_dde.c |   25
> +++++++++++++++++--------
> >  1 files changed, 17 insertions(+), 8 deletions(-)
> >
> Hi Mikey,
> 
> There's also a test failure on localized Windows (see
> http://test.winehq.org/data/8a17a028b73d1ccc5b7ebaa9e135a1bfbf159615/2000_w2k-sp4-fr/shell32:progman_dde.html
> for example) because the window doesn't have the expected
> title.
> 
> I tried to skip the ShowGroupTest call at line 510 (before
> your patch)
> but then the next call fails. Is there a better solution
> than skipping
> those two calls ? If not I'll just send a skipping patch.
> 
> 
> -- 
> Nicolas Le Cam
> 


      



More information about the wine-devel mailing list