Le 02/02/2011 05:54, Phil Steitz a écrit : > On 2/1/11 2:06 PM, Luc Maisonobe wrote: >> Le 01/02/2011 18:22, Jörg Schaible a écrit : >>> Hi, >>> >>> Phil Steitz wrote: >>> >>>> We are in process of redesigning our exceptions hierarchy in [math] >>>> and we could use some input / perspective from other Commons >>>> community members. Thanks in advance for your feedback and perspective. >>>> >>>> The most recent attempt at agreeing on design principles is [1] and >>>> I have tried to document the points of agreement in a section that I >>>> just added to the [math] Developer's Guide [2] (it will take a few >>>> hours for this to make it to the live site. The source xdoc is in >>>> trunk/src/site/xdoc/developers.xml) >>>> >>>> The JIRA issue [3] referenced in [1] calls out a specific point on >>>> which we are trying to gain consensus in a specific example. >>>> >>>> The currently defined exceptions in [math] can be found in the >>>> top-level package and .exceptions. Those in the top-level have at >>>> this point been deprecated. >>> Has anyone of you considered the usage of an exception context? We >>> introduce >>> this with lang3 (http://commons.apache.org/lang/api-3.0- >>> beta/org/apache/commons/lang3/exception/ExceptionContext.html) and it >>> basically adds a map with key/value pairs to the exception itself. >>> >>> This concept has worked out quite nicely especially in situations where >>> code >>> can nest deeply in conjunction with user-provided code that make calls in >>> the core functionality again. >>> >>> The nice part is that you can add information at each position that can >>> provide useful information (assuming MathException implements this >>> interface): >> We didn't consider that :-( >> It seems very interesting to me. >> >> What do other think about it ? >> > Thanks, Jörg! > > Very nice idea. I can see added value implementing the interface; > but unfortunately I don't see it replacing either the localization / > message framework or the hierarchy. For reasons that others have > stated, I think that we do need a simple hierarchy. We also need > messages that convey more than data values and I am concerned that > making sense out of key/value pairs in messages would require > brittle convention-maintaining. Have a look at the messages that we > maintain today: > http://s.apache.org/NNd > It could be, though, that the embedded machinery might be useful and > allowing users to add data to exception messages is a powerful feature.
It seems the thread asking for help on the exception API design is going to be fruitful, and it starts well with interesting ideas. I guess some of these ideas will change again our view and we will converge (hopefully not throwing an exception ourselves ...) to a stable design for 3.0. It seems to me that the switch to unchecked exceptions is supported by almost all participants to this thread, so this part is probably already stabilized. I doubt we can do anything about it for 2.2 and waiting first for the rest of the discussion to stabilize (no hierarchy/small hierarchy/large hierarchy, specific getters/general context map ...) would push 2.2 too far. I would like to freeze 2.2 as it is now in the repository and get it out. What do you think ? Luc > > Phil > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org > For additional commands, e-mail: dev-h...@commons.apache.org > > --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
