Re: Oracle For Fun

Daniel Morgan <damorgan_at_x.washington.edu> wrote in message news:<1075568670.136040_at_yasure>...
> Tony wrote:
>
> > "Niall Litchfield" <niall.litchfield_at_dial.pipex.com> wrote in message news:<401ae429$0$10057$cc9e4d1f_at_news.dial.pipex.com>...
> >
> >>"Tony" <andrewst_at_onetel.net.uk> wrote in message
> >>news:c0e3f26e.0401301236.2b11876a_at_posting.google.com...
> >>
> >>>Why would a production environment require redundant error handling
> >>>and variable declarations? ;o)
> >>
> >>Variable declarations for readability essentially.
> >>A do nothing error handler for when the code gets changed to do something
> >>that might actually throw errors. People always exclude error handling from
> >>their code. It is almost always a mistake.
> >
> >
> > I tend to the opposite view: some people put too much error handling
> > in their code, sometimes resulting in unwanted behaviour. Without
> > explicit error handling, the worst that happens is you get a terse
> > error message; with bad error handling, you may never know there was
> > an error.
>
> I think you're taking a bit too much of a "position". Lets look at this
> from a different perspective. Is there any excuse for putting any code
> in a database that does not have an exception handler?
>
> I can think of only one ... a procedure that logs errors in which case
> you want the app to blow up if it ever fails because its failure would
> mean that errors are not being logged.
>
> Having established, I hope, that there is only a few very rare cases
> where an exception handler isn't required ... the question of what
> constitutes too much or too little becomes academic.
>
> The amount of exeption handling should be complete enough to trap all
> possible errors, log them if appropriate, and either continue processing
> or exit gracefully.
>
> But from my experience ... I've seen thousands of cases of none ...
> never a case of too much. Your mileage may vary.

Speaking of mileage, the CA DMV site had a case of too much. They had a web app that would allow you to check for smog tests by putting in a license plate. Of course, it was just passing through, so you could put in wildcards, if you didn't put in one that would return so much that it would timeout. Errors, including timeouts, were logged to another database. Naturally, if the other database were down, it would blab to the users browser. They appear to have upgraded since then. Now nothing at all works, stoopid java code too slow. Amazing what you can learn with a browser.

jg

--
@home.com is bogus.
http://www.signonsandiego.com/news/metro/crowe/20040202-1350-tuite-escape.html