Re: Tin Man Design Question 10 - more thoughts

[David_Marston@lotus.com]

>> is me, earlier
> is Ken

>>"The most flexibility is achieved by requiring a separate file that has
>>name/type/value entries, like an old .INI file with enhancements. It
>>would be up to the test lab, possibly with OASIS guidance, to determine
>>how to extract the values from the file and present them to each XSLT
>>processor." We'll presume a file named <Source>.ini for the time being.

>Can we use an XML file for this information?

Maybe. The name and type fields will not be a problem; the value might
be. Xalan-Java 1.x allowed the full expression syntax on the command
line, which imposes some quoting requirements and implies that many
punctuation marks could appear, including < and >. Therefore, use of
CDATA sections becomes likely.

If we can use XML syntax, we don't have to have a separate file. We
could put everything inside the <Param-set> element in the catalog.
This possibility should be carefully examined by those who will build
the part of the system that runs a test case based on catalog data.
Would you rather have a separate XML file that can be transformed (by
XSLT, of course) into a batch file or shell script, or could you work
equally well by transforming some elements from the catalog? I think
the former is likely to be much easier.
.................David Marston