Re: Failed gui test:NSView_autoresize_and_rounding WAS: gnustep compiled with clang and gworkspace problem

[Fred Kiefer]

On 30.10.2011 14:07, Sebastian Reitenbach wrote:
> Hi,
>
> since -base tests are fine now for me, I'm back to -gui: The following tests 
> are still failing for me on OpenBSD i386 5.0 -current:
>
>
> Testing lazy_copy.m...
> Running gui/NSPasteboard/lazy_copy.m...
> which happens to fail randomly.
>
> Testing setDelegate_reload.m...
> Running gui/NSSavePanel/setDelegate_reload.m...
> Failed test:     browser initially contains all files
> Failed test:     browser is reloaded after -setDelegate:
> Failed test:     browser contains all files after resetting delegate
> Failed test:     browser is reloaded after -setDelegate: (2)
> Completed file:  setDelegate_reload.m

The most likely reason for these tests to fail is that the .svn 
directory is missing. We should rewrite the test in a way that does not 
rely on the fact that the code includes the .svn directories. The will 
be gone in a while anyway when everybody switches to svn 1.7.
And of course the test "browser contains all files after resetting 
delegate" fails on all systems.

> Testing NSView_autoresize_and_rounding.m...
> Running gui/NSView/NSView_autoresize_and_rounding.m...
> Failed test:     NSView autoresizing works
> Passed test:     NSView autoresizing corner cases work
> Failed test:     NSView autoresize rounding works
> (1) expected frame (10 10)+(10 10), got (10 10)+(9 10)
> (1) expected frame (10 10)+(10 10), got (9 9)+(9 9)
> Completed file:  NSView_autoresize_and_rounding.m

These tests should work. Here you could try to add some NSLog statements 
to the code in NSView to find out what is actually going on. I remember 
that when you looked at that code in gdb you got unlikely results.


> Testing NSView_bounds_scale.m...
> Running gui/NSView/NSView_bounds_scale.m...
> expected bounds rotation 30 got 30.9153
> (2) expected bounds (39.9801 104.211)+(66.2393 85.2544), got (40.372 
> 105.364)+(66.8473 84.9812)
> expected bounds matrix (2.93813 0.614059 -1.69633 1.06358 104.31 -131.062) got 
> (2.96586 0.574951 -1.71234 1.07291 104.31 -131.062)
> (2) expected bounds (39.9801 104.211)+(66.2393 85.2544), got (40.372 
> 105.364)+(66.8473 84.9812)
> expected bounds rotation 30 got 30.9153
> Completed file:  NSView_bounds_scale.m

These errors are currently to be expected. I spend quite some time on 
that problem this summer and have now given up to reproduce what Apple 
does here. To me it seems wrong what they are doing and inconsistent too.