Prob depends on situation. For us it works.
E.g. you have a SQL Transcation which affected 10 tables,
Now you would then need to
go from Backend --> SQL 10 times putting locks on a ****load of places while moving between backend and sql, instead of having it in one stored proc thereby limiting the amount of posts (and potentially lag) between app and database .
If there is a problem between backend and SQL - everything rolls back - user has to enter everything again - probably even not being informed of error, etc etc etc,
where with an SP you'd have the process complete irrespective of whether it will ever hit backend code again - i.e. if the user logs on again after connection issues he has a higher likelihood that the transaction completed and he wont have to do the transaction again
OR
build this baby up in Dynamic SQL? Holy smoke no. I rather avoid dynamic SQL than SP's even though both have uses.
but at least we get the chance to rethink our methods and see the opinion of others, so this thread has actual value.
On the 2004
article
1) very debatable
2) i
debug SP's every day
3) i get plenty feedback from SP's - IDs that were created, rollback info, info on exactly what went wrong
4) pass objects to SP's every day -
SQL 2008
5) If the DBA wont allow you access, well then yeah. All IDEs give you access to view source even if only read only
but yeah, to each his own, some places it will prob work, other places prob not.