W3C home > Mailing lists > Public > public-rif-wg@w3.org > September 2008

Re: the Language problem with Test Cases

From: Stella Mitchell <cleo@us.ibm.com>
Date: Thu, 18 Sep 2008 22:10:29 -0400
To: Sandro Hawke <sandro@w3.org>
Cc: public-rif-wg@w3.org, public-rif-wg-request@w3.org
Message-ID: <OF23ED3888.62AAEB05-ON852574C9.000BA08B-852574C9.000BF257@us.ibm.com>
This sounds ok. In practical terms, it  seems like this means
that in the near term, we'll use the syntax that  Hassan's planned
translator takes as input  (or write directly in xml.)

In the proposal below, I think you should also mention
the specification (not just implementation) of the mapping
from the OS->XML (OS=other syntax), and address under
what conditions review would occur. I think that the 
people  who vote to approve a test case will
need to either:
   1.  review the XML
   2. review the mapping from OS to RIF,  and review 
       the OS representation of the test,  and trust  the
       implementation of the translator.
For test cases that don't come with an associated  BLD-PS
representation (either because they were written directly in
XML or because OS != PS),  it  would be good if at some point
we could provide an XML->PS translator (as Christian 
brought up the idea of at  the telecon).  I'm assuming that 
a BLD-PS representation of the test case would be helpful to
people (implementers, testers) who are not familiar with OS
and who want to relate the syntax  in the test case to the semantics
in the BLD spec, without having to read the OS specification
(and there could be 20 OSs),  or the XML version of the test,
and then do various mental translations to get from os to xml
 to ps to semantics. But I'm not sure - maybe most  will just think 
in terms of  xml -> semantics.


Sandro Hawke <sandro@w3.org> 
Sent by: public-rif-wg-request@w3.org
09/18/2008 11:34 AM


the Language problem with Test Cases

Chris, Christian, and I hammered out something that looks like it allows
us to proceed with authoring test cases.  There are several challenges
we're trying to circumvent:

      - Authoring RIF XML by hand is very hard (and error-prone)
      - Reading RIF XML is very hard
      - We have, as yet, no software to translate between readable 
        syntaxes and RIF XML
      - We have no stable, fully-defined, usable Presentation Syntax

We don't want to wait for solutions to any of the above problems, but we
want to continue working on test cases.   So, here's the proposal for

     1. Eventually, every RIF test case will have the inputs be
        expressed in XML. This XML is what we expect machines to use in
        running the test cases. 

     2. Most RIF test cases will also be expressed in some human
        syntax.  For this part, we are only interested in syntaxes for
        which a RIF translator implementation is promised.

     3. At some point, the implementations SHOULD be made available so
        that the XML can be generated or tested.  If it is not, the XML
        must be generated and checked by hand (or the test case will
        have to be withdrawn).

This seems to us like the best way to proceed at this point.   It'll be
on the agenda for Tuesday.

       -- Sandro
Received on Friday, 19 September 2008 02:11:11 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:47:52 UTC