- From: Dave Reynolds <der@hplb.hpl.hp.com>
- Date: Tue, 19 Jun 2007 15:51:42 +0100
- To: Philippe Bonnard <pbonnard@ilog.fr>
- CC: RIF WG <public-rif-wg@w3.org>
Philippe Bonnard wrote: > > > Hello, > > > > I have the following concerns about the list of buildins that may be > included in RIF core. > > · Could it be possible to limit to functions with an easy > implementation in main language? For example, the match function may be > tricky to implement in a language where there is no libraries concerning > regexp, > > · Up-to-now, I have not the time to check that the selected > functions could be quickly implemented in Java, especially for the date > types. Is it possible to have more time to perform this task? Xerces includes an xpath/regex package in Java under the Apache license. I have no information on how functional or conformant that is or how easy to use separate from the rest of Xerces but it is maybe some sort of data point. > · Those functions are perfect to operate against an XML model. > However, most of our PR ruleset are not compliant with such a model > since they use methods coming from Java libraries or user's libraries. I > don't think we will be able to bind those methods and data on XML and > XPath functions. Furthermore, even if it exists very similar methods in > the Java library, some of them have a slightly different semantic. Take > for example the '+' operator of Java which differs with 'op:numeric-add' > from the way the overflow is processed. I thought op:numeric-add overflow was only an issue for xsd:float and xsd:double, neither of which are currently in RIF core. [OK, that's a slightly weasel-like comment since it immediately begs the question of why we don't have those in RIF and pretty soon we'll find ourselves deciding we do want floating point and your comment would then be right.] Cheers, Dave -- Hewlett-Packard Limited Registered Office: Cain Road, Bracknell, Berks RG12 1HN Registered No: 690597 England
Received on Tuesday, 19 June 2007 14:51:57 UTC