infoset based comparison

[Kirill Gavrylyuk <kirillg@microsoft.com>]

Sorry being late on this.
Attached is a sample suite of Microsoft xsl:number tests, that uses
infoset-based comparison for xml and text output methods. For xml files
we use I'-transform which describes  subset of the document's infoset.
It contains:

Number-formatting\*.* 	-	folder with xml and xsl files + infoset
transform of xml input and expected output files (prefixed  as i-
files).
control.xml			-	control file made along the
lines of David's last iron man edition of the document
infosetgen.xsl			-	xslt that produces I' image for
xml files
makeexpect.js			-	sample script that produces
infoset images of xml-input files and expected output files (both xml
and text, using msxml3).
test.js				-	runs tests against msxml3

Problems that were not solved:
1) How to deal with encodings other then UTF-8? Proposed idea is to put
encoding attribute on the I' files 
2) HTML files
3) need to add I'-image to scenario element in control file.

I'll follow up on the David's mail on Iron man for other suggestions.
 <<msft-number.zip>> 
Please try this and tell me what do you think.
Thank you!

msft-number.zip