From: Eric B. <er...@go...> - 2008-01-02 19:49:39
|
Colin Paul Adams wrote: > I have factored out the core code for the XSLT functions > format-date(), format-time() and format-date-time() and moved it into > the time library (this introduces dependencies for the time library on > the math and regexp libraries, but I think this should not be > considered a problem), and added a new function named xslt_formatted > to each of DT_DATE, DT_TIME and DT_DATE_TIME. Introducing new dependencies should always be considered as a problem and should be considered with care. My first question is: what does this refactoring bring to the users? I can see that people who don't care about XSLT and XPath will have more classes to compile. Will there be people who will use these new functionalities in the date/time library and not the xslt/xpath libraries? I don't really have time now to review this code that have been added to the date/time library, but just looking at the names of the new classes seems very suspicious to me (DT_XSLT_..., DT_XPATH_...). If it is likely that people who will use these new functionalities will also use the xslt/xpath libraries, then what about having all that in a cluster 'time' in these libraries? Also, I wonder what is peculiar to date/time in class DT_STRING_ROUTINES. Likewise with DT_XSLT_NUMBER_ROUTINES. -- Eric Bezault mailto:er...@go... http://www.gobosoft.com |