[jdom-interest] (no subject)

Jason Hunter jhunter at acm.org
Mon Jun 2 14:20:41 PDT 2003


> I did'nt say that but I bet on the API 2 years ago. At this time, it was'nt supposed to take more than 3 years for
> having something usable and stable.

I fully agree with that.  :-)

> I do.
> But, for example, within the TODO coming with Beta 9, there isn't any reference to the new interfaces as well as the
> method renaming. Is it related to "Look at Brad's experimental tree for good ideas to integrate"?

We've had a long item on what we should do for a Node interface.  Node 
split into Parent/Child since that was the best breakout without useless 
methods.  You can argue there should be a parent Node interface, but I 
don't (yet?) see what would go in it or what advantage it would serve.

> Another example is about making flushCharacters(String) private, while it was specifically designed to be overloaded.
> Hopefully, Laurent detected that :-)

Yep.

>>>>I know we need to break some things for 1.0<<<
> 
> What is the objective? 

The objective is to have a strong 1.0 release.  We're going to live with 
1.0 for a long time, so it should be right.

> BTW, about >>>IOException are being to be thrown since Beta 9 instead of only JDOM exception.
> Yep, another place where we broke things but it was the right thing.<<<
> I disagree with that and prefer the notion of exception wrapper, where a JDOM Exception owns the IOException (which may
> also own another exception...). For my opinion, a library should only throw exceptions whose inherit from a unique base,
> and do not throw exception that rely on the implementation. Else, it breaks the library isolation.
> IOException is a good example for that: maybe that your stream is going through the network, but you did'nt have a
> direct access to any network exception. Using the parent exception paradigm, you can get it.
> But that discussion is now closed...

Did you comment at the time?

-jh-




More information about the jdom-interest mailing list