From: David Naseby <david.naseby@...>
Date: 2002-12-06T11:43:29+09:00
Subject: Re: on error resume next

>-----Original Message-----
>From: Shannon Fang [mailto:xrfang@hotmail.com]
>
>> > I tried to do the following: (peudo-code)
>> >
>> > def test
>> >        begin
>> > 	conn=adodb.connection
>> > 	conn.open ...
>> > 	if conn!=nil then
>> > 		#add data to db
>> > 	else
>> > 		#add data to file
>> > 	end
>> >        rescue DBError
>> > 	conn=nil
>> > 	resume next
>> >        end
>> > end
>> 
>> i don't quite understand this...  the next WHAT?  where is 
>the implied loop
>> supposed to be?
>
>The benefit of resume next is that it will catch all errors. Here, I
>expect error to happen on the conn.open statement., so if error happen,
>set conn=nil.
>
This thread seems so wrong on so many levels. Wishing for VB-like 'error
handling' is.. well.. its quite beyond my comprehension. If you really want
to achieve the implied behaviour of the piece of pseudo-code above, try:
def test
  begin
    conn=adodb.connection
    conn.open ...
  rescue ADatabaseError
    conn = nil
  end
  if conn.nil? then
    #add data to file
  else
    #add data to db
  end
ensure
  conn.Close unless conn.nil?
end

But I think you really have to ask yourself, will this give me a good system
state if some random error can throw my (logging/ reporting/ whatever DB is
supposed to hold) to file at any random moment? How do you later reconcile
the DB and file? If this is supposed to be a program level switch, write a
DoWhateverImDoingToStoreData class that can polymorphically write to file or
database depending on a factory level switch, and pick it at runtime - to
switch mid program as the result of an error seems like folly.

YMMV

David Naseby