My preference: Abstract classes for everything except Listeners. There is a well defined JavaBeans pattern for these.
We can work the rest out closer to the date. Probably the biggest overriding design philosophy for SWT is "Less is more". Give only API that is really needed. Reuse names and concepts where it makes sense so the consumer of your API has less to learn. Define only one way to do things. The list goes on and on and gets more preachy so I'll stop here. "David J. Orme" <[EMAIL PROTECTED]> Sent by: [EMAIL PROTECTED] 01/19/2007 03:14 PM Please respond to Nebula Dev <[email protected]> To [email protected] cc Subject [nebula-dev] API code review before M5? Thanks! Three questions: 0) (to the other committers) Who in Nebula is interested in participating? I'm fine with using email as that makes it easier to have a deeper conversation in a time-decoupled manner. 1) What kind of time frame works for Steve? For everyone else? Or maybe it's easier for everyone if individual Nebula commiters work separately with the SWT team (in which case, question 0 doesn't matter)? 2) Steve: Is there a list of things we should think about before we start the review? Here's what I'm thinking of right now: - JavaDoc completed for all API classes and methods. - Obvious refactorings completed. API should obey DRY and the various concerns should be defined where they're used, etc. - Attention paid to the points raised about good SWT control design on the Nebula web site. - One question: SWT seems to prefer to define listener APIs using interfaces rather than abstract classes, and to provide Adapter classes for folks that don't need to implement a full interface, consistent with AWT. Others API authors I've spoken with prefer to define API using abstract classes rather than interfaces. Their reason is that you can evolve an abstract class without breaking clients but if you add a method to an interface, you break everybody. What is your preference? Is it a higher priority for new API to be consistent with AWT and the rest of SWT or to be more evolvable going forward? Do you care? Is there anything else I'm missing? Thanks in advance. :-) Regards, Dave ----- Original Message ----- Date: Fri, 19 Jan 2007 09:47:10 -0500 From: Steve Northover <[EMAIL PROTECTED]> Subject: Re: [nebula-dev] API code review before M5? To: Nebula Dev <[email protected]> Message-ID: <[EMAIL PROTECTED]> Content-Type: text/plain; charset="us-ascii" I'm into it. _______________________________________________ nebula-dev mailing list [email protected] https://dev.eclipse.org/mailman/listinfo/nebula-dev
_______________________________________________ nebula-dev mailing list [email protected] https://dev.eclipse.org/mailman/listinfo/nebula-dev
