Re: My Oracle::OCI presentation from the Database Discovery Geek Cruise (in PowerPoint)

[email protected] (Tim Bunce) Mon, 9 Jul 2001 19:00:08 +0100
Newsgroups perl.dbi.oracle-oci
Organization Paul Ingram Group, Software Systems, +44 1 483 862800
Message-ID <[email protected]>
On Mon, Jul 09, 2001 at 09:12:16AM -0500, Blackstone, J. David wrote:
>   I notice you proposed the ocistrlen($string) function to return ($string,
> length($string).  Wouldn't it be better for one of the "small set of rules
> for changing OCI functions to Oracle::OCI functions" you mentioned to be
> "functions that take a string, length pair will instead just take the
> string" and have the glue code substitute in the two values?  That would
> seem to make it more Perl like.

The function has been implemented as

	oci_buf_len($string)
	oci_buf_len($string, $max_buf_len)
	oci_buf_len($string, $max_buf_len, \$alt_out_len)

A key feature of Oracle::OCI is that the code is almost totally
generated from the Oracle header files. There's no reliable way
to tell which OCI function arguments are buffer+length pairs.
Also it would add 'distance' between the raw OCI calls and the perl
versions.  I'm trying to keep that distance as small as possible.

Try it out and see how it works in practice.

Tim.