Re: [Java-player-developers] Need to propagate exceptions, not hide them.

[Radu B. R. <ru...@cs...>]

Hey Gareth,

Thanks. I know that all Javaclient users will appreciate this. :)

I am slowly preparing to start coding the diffs for handling Player 2.0. 
In the meantime, I have coded a new interface (RFID) for Player and the 
appropriate code for Javaclient and the C/C++ client. I will post the 
patches here and on my web page as soon as possible, and with a little 
bit of luck, I shall be able to convince Brian to merge them with the 
Player2 repo, once it's stabilized. Right now I am handling two brands 
of RFID readers, but I'm sure that once the code is out, people will add 
support for their own.

If anyone here is working with RFID and has any requests on what this 
interface should contain, please speak now. For beginning, I'm adding 
Select and then Read/Write support to the tags, but we might make use of 
encryption functions later too, if anybody needs them (Inside readers 
have a bunch of techniques for that).

PS. Gareth, you might want to CC e-mails related to this subject to 
java-player-users@sf, since I know for a fact that some of our users 
aren't subscribed to -developers, and vice-versa. Thanks.

Best regards,
Radu.

-- 
| Radu Bogdan Rusu | http://rbrusu.com/
| http://www9.cs.tum.edu/people/rusu/
| Intelligent Autonomous Systems
| Technische Universitaet Muenchen


Gareth Randall wrote:

>Hi Radu,
>
>I'll see what I can do over the next 2 weeks, and send CVS diffs :-)
>
>Does anyone on the list have views as to whether exceptions arising
>from calling javaclient methods should be checked or unchecked
>exceptions?
>
>Most of the issues that are being caught look rather serious -
>sufficiently serious that most callers of the code will not want to try
>to handle them. E.g. if something is serious enough for you to be
>calling System.exit(1) then it's unlikely that a client will be able to
>recover !
>
>This suggests that they should be unchecked (i.e. extend
>RuntimeException). If a caller really wants to then they can catch
>them, but otherwise they need not be burdened with the extra code.
>
>A further argument for unchecked comes from SQLException. One of the
>(many) criticisms of "java.sql.SQLException" is that most database
>errors are simply unrecoverable without human intervention, and that
>there is no point in a caller being forced to catch them. This argument
>seems to apply to javaclient.
>
>
>Unchecked is best unless someone can think of a compelling reason
>otherwise.
>
>Yours,
>
>Gareth Randall
>  
>