[Mono-devel-list] Re: Analysis of W3C regressions

[Atsushi Eno]

I got very different results from that W3C tests.

As long as I tried System.XML/Test/System.Xml/standalone_tests
(which is stable, unchanged) there was no difference reported.

I need concrete number unlike just saying "there are regressions".
For now what I got from r41840 is

Total:2162
Performed:2162
Passed:1424
Failed:709
Regressions:488
Fixed:66

and from the latest 45852

*********
Total:2162
Performed:2162
Passed:1865
Failed:268
Regressions:0
Fixed:10

Of course I tested both on the same environment and the same
runtime (version and configuration).

Atsushi Eno

Andrew Skiba wrote:
> Hello.
> 
> I started to analyze the list of regressions that are found with W3C 
> testsuite. They are listed in fixme.lst file in svn.
> 
> I found that undoing revision 41840 fixes 20 regressions from the list 
> (although breaks one test).
> 
> I attach here the patch that you can apply to the latest System.XML to 
> fix those 20 regressions. Please notice, it's not my patch, it's just a 
> revert of a part of revision 41840. I made it so it's easy to verify 
> these results.
> 
> Before the patch there are 0 regressions, and 10 fixes, after the 
> attached patch there is 1 regressions and 30 fixes.
> 
> Please tell me what you think.
> 
> Cheers,
> Andrew.