W3C home > Mailing lists > Public > uri@w3.org > April 2004

Re: test cases and implementation reports (UPDATE)

From: Graham Klyne <GK@ninebynine.org>
Date: Wed, 21 Apr 2004 17:55:49 +0100
Message-Id: <>
To: uri@w3.org

Further to my previous report, my URI handler now passes all my test cases 
(after changing two normalization tests to correspond to the final .. 
normalization specification):
URITest> main
Cases: 515  Tried: 515  Errors: 0  Failures: 0


One of my test cases is a corner case whose correct outcome is not entirely 
clear to me from the specification.  It's "Normalize28" in my test 
descriptions.  I now have:

Uri:               foo:a/b/../.././../../e
Path-normalized:   foo:e

I've updated the files referenced via:


(previous message follows...)

In the process of revisiting my implementation, I've extracted all of my 
test cases (which include those in the spec, and many more) into an 
external description using RDF [1], and developed a new test harness to be 
driven by the test descriptions rather than by hand-coded test cases.  I 
think it would be possible (and not too hard) to generate test code for a 
variety of different platforms from the RDF data.

At this time, I still have to revise my code to bring it in line with the 
final decision on handling of leading dot segments.  Otherwise, all tests 
are passed OK [2].



[2]  URITest> main
### Failure in: 195
expected: "http://a/g"
but got: "http://a/../g"
### Failure in: 197
expected: "http://a/g"
but got: "http://a/../../g"
### Failure in: 198
expected: "http://a/g"
but got: "http://a/../g"
### Failure in: 235
expected: "foo:"
but got: "foo:.."
Cases: 513  Tried: 513  Errors: 0  Failures: 4


Graham Klyne
For email:
Received on Wednesday, 21 April 2004 13:00:23 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:25:07 UTC