Spam: Re: [Mono-devel-list] minor xslttest makefile patch

[Atsushi Eno]

Hi Andrews,

Here's a couple of comments on those "regressions".

 > BVTs_bvt083

I believe this is a bug in the testcase. In the corresponding
xsl:for-each, there is a text node (which looks like a whitespace
node which is however preserved by xsl:space) and thus xsl:sort
comes after "template" and thus it should be rejected.

 > ConflictResolution__77902
 > Elements__89171
 > Elements__89177
 > Include__77736
 > Include_Include_IncludedStylesheetShouldHaveDifferentBaseUri
 > Include_RelUriTest1
 > Output__84011
 > Text__78309

Apparently your sed script needs more lovely fix.

Can you make sure that there is no difference between the old
catalog.xml which you removed from svn and what you generate
today via the sed script?

 > FormatNumber_FormatNumberWithPattern.00
 > FormatNumber_FormatNumberWithPattern.000
 > XSLTFunctions_RoundTripNumber_UsingStringFn

Ok, there seems a bug in either DecimalFormat.cs or mscorlib number
formatting. I'll try to make a repro on custom NumberFormatInfo.

 > Output_HtmlOutputWithAmpersandCurlyBracket
 > Output_HtmlWithTwoAdjacentEscapedAmpersands

There seems a bug maybe on determining output method, or just a
bug in HTML outputter (I doubt former one).

 > Sorting__78191
 > Sorting__78286

There is a difference at string comparison in INT output element
(yes, it is string text comparison). Text comparison depends on
collation and now without ICU it just compare characters by
codepoint. With full collation support, '-' is kinda ignored
(unless we use StringSort option when comparing strings) and
thus the output should be like the reference output.

 > XSLTFunctions__emptyParameters

This is also rejected in MS.NET.

We should try to compile the stylesheet and expect error here.
I have no idea why it worked fine before and it does not today.

Atsushi Eno