[CDBI] mysql compress support

John Siracusa siracusa at mindspring.com
Mon Sep 19 18:18:13 BST 2005


On Tue Aug 30 15:04:01 Matt S Trout wrote:
> Rose::DB has a reasonable interface and is very fast, but it's making the
> same SQL-liberally-sprinkled-in-the-code mistakes that the Tangram project
> is currently working hard to refactor out (and making fantastic progress too
> - you should drop in and see how they're getting on as well).
>
> For example (from Rose::DB::Object::MetaData) -
> 
> sub insert_sql
> {
>   my($self) = shift;
> 
>   no warnings;
>   return $self->{'insert_sql'} ||=
>     'INSERT INTO ' . $self->fq_table_sql . "\n(\n" .
>     join(",\n", map { "  $_" } $self->column_names) .
>     "\n)\nVALUES\n(\n" . join(",\n", map { "  ?" } $self->column_names) .
>     "\n)";
> }

That's not a "mistake" unless it prevents RDBO from achieving its design
goals.  I don't intend to support anything but SQL-based databases via DBI.
OTOH, speed is an important design goal, for which I've traded a lot--but
only in the internals, not in the public or "protected" API (of which the
above code is not part, BTW).  I suffer so you don't have to(tm) ;)

(sorry for the late reply, I just joined :)
-John






More information about the ClassDBI mailing list