- From: Andrew Clover <and-w3@doxdesk.com>
- Date: Fri, 19 Dec 2003 22:04:24 +0000
- To: www-dom-ts@w3.org
Curt Arnold <carnold@houston.rr.com> wrote: > Does the terminal whitespace cause the test to fail? Yes, because the next-sibling-of-next-sibling tested is whitespace. (Similarly another test uses a previous-sibling test to check INSERT_BEFORE.) > I think I know a way that this test could be done on Java without having > an external resource. Pretty complicated, so it isn't going to be on > the top of my list until I just need a challenge. [...] Ouch. Yeah, that seems a bit excessive! I don't think an external resource is necessarily such a bad thing, if made available on a known server and optional (so one could still use localhost instead if the external server were not contactable). > Cool. Hopefully, it didn't overwrite your home page. Heh. (No, I haven't got a proper HTTP PUT server hanging around, so I just hacked up a trivial CGI to respond 201 to any request.) > SystemId1 attempts to write to a file: URL. I think it > represents a reasonable expectation, but might overreach what is > actually in the spec. I think this is more than reasonable, yes. > If don't have priviledges to write a temporary file, SystemId1 can fail. It could perhaps be the test runtime's responsibility to check that a createTempSomethingURI refers to a resource that can successfully be written to before calling the tested implementation. > I'd would like it if the names could be synchronized that is renaming > nodeArg to n Me too. I guess no project in the world other than TS actually depends on the exact argument names given in the spec, but this is still desirable from a consistency point of view. (Currently I've had to special-case the TSML interpreter for acceptNode, as well as XPath's evaluate.) -- Andrew Clover mailto:and@doxdesk.com http://www.doxdesk.com/
Received on Friday, 19 December 2003 17:26:56 UTC