[Geotools-devel] Proposal for DataSourceException to extend IOException

[Chris H. <ch...@op...>]

IanS and I have been working through some of his DataSource ideas today, 
and one area that we've hit upon is our DataSourceExceptions, a 
traditional point of contention between jdbc and file datasources, as jdbc 
always throws SQLExceptions, while files throw IOExceptions.  

Throwing IOExceptions seems to make a bit more sense, as not all our 
datasources are SQL-based, but it also seems a bit out of place to have an 
error in resultSet.close() throw an IOException.

So our new proposal is to have DataSourceException sub-class IOException.  
We'll still be able to wrap other exceptions, even though IOException 
doesn't allow this in it's constructor, as Ian has come up with a way to 
work around it.  

Doing things this way will allow clients to just catch IOExceptions, 
instead of having to catch both, and is just a bit cleaner way of doing 
things.  rmi Remote Exceptions and Corba Exceptions also sub class 
IOException.  Also, see http://www.artima.com/intv/solid3.html for a 
mention of IOExceptions.

IanS and I are both +1 on this proposal.


	
	Chris