On Jan 23, 2008 2:07 PM, Ted Zlatanov <[EMAIL PROTECTED]> wrote:
> I found one of the problems is that you forcibly set RaiseError to 1 in
> a lot of places, in Rose::DB and Rose::DB::Object. Why override the
> connect_options?
>
> My code is working with do_transaction for now, but begin_work/commit
> should really be shown with an eval in the Rose::DB SYNOPSIS. I got the
> impression they would not die from the Rose::DB docs, so their
> exceptions were killing an outer eval, and rollback() was not called...
> I'd show you the code but it's not interesting or pretty.
The only instances I can think of off the top of my head are
local()ized changes so that I can use eval blocks internally for error
handling. For example:
sub whatever
{
...
eval
{
local $dbh->{'RaiseError'} = 1;
# do stuff
};
if($@) { ... } # handler error
else { ... } # or success
}
If there's an instance where RaiseError is altered outside and eval
block and/or not changed back before a sub returns, it's probably a
bug, so send me any cases you've found that demonstrate this.
> I ran with DBI_TRACE set high and checked that the DBD commit was only
> called as many times as I specified. So I'm using transactions to
> insert the rows, and RDBO is just not able to handle the load for my
> particular task. With COPY FROM, I can insert about 10-20 times faster
> than RDBO, so I'll have to stick with that. I'll mimic the RDBO SQL
> statements and just run them I guess...
Did you try plain DBI? I suspect even that will be 10-15 times slower
than COPY FROM. Native bulk loading is always going to win, by a lot.
-John
-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Rose-db-object mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/rose-db-object
