[Chicago-talk] blessing horrors

[Jim Thomason]

>Man, that's the kind of stuff that makes me lose
>sleep. To wax a bit, part of what I've always enjoyed
>about this junk is that the answer is out there if
you >dig deep enough (I'm a near-obsessive debugger)
and >some of what has motivated/rewarded such
'dedication' >is that it almost always leads to the
discovery of a >bigger flaw.  There aren't often
'little anomolies' >but little hints to bigger
problems.
 
I agree...up to a point. In this case, I did dig a lot
deeper and discovered a bigger flaw instead of a
little anomoly. 

The problem manifested itself when I was trying to use
a module that governed database transactions. Has a
flag to determine if they're on or not. It worked on
the command line, but not through apache. Initial
investigation yielded a "solution" that worked.

It was the deeper digging that caused me to learn
about all of this nonsense.

Here, I'd say that the bigger bug was the fact that my
code was expecting the package to never start with
"main::" or "::", and it turns out that that's not a
reasonable assumption to make. So that snippet of code
was corrected.

Sure, I could go ahead and dig through the other hunks
of code to see where the 'main::' is being generated,
but it's really not terribly important. My code should
handle it (there may be some other point in the future
where a 'main::package' could be handed in by someone,
and I need to account for it), and it does. The fact
that it's generated at all? Sure, it's annoying, but
not necessarily worth the time and effort to fix,
since the issue is now handled by the more important
fix that the digging unearthed.

I file that 'main::' generation under the list of
fringe cases that the core should account for. Sure,
it'd be nice to isolate, but there's more important
work to do in this case.

-Jim......


	
		
__________________________________
Do you Yahoo!?
Yahoo! Tax Center - File online by April 15th
http://taxes.yahoo.com/filing.html