On Tuesday 09 August 2005 21:33, Garrett Rooney wrote:
> Nick Kew wrote:
> > Anyway, as I said, a placeholder now fixes that.
>
> Out of curiosity, is there any reason we can't just forward declare the
> struct and access the internals via some accessor functions?

Hmmm, fair point.

There's one definite stumbling block there: the varargs APIs.
Can the vararg functions (specifically, the pvquery and pvselect
functions) be expressed as apr_dbd #defines, like the rest of
the calls?  Anyone?

> I mean 
> it's a database library, so it's going to be doing a lot of I/O, I can't
> imagine the overhead of the function calls would be that big of a
> problem in this case...  That would allow us to extend it indefinitely
> in the future, as opposed to the limited extensibility provided by
> padding out the struct in the public header file.

Indeed.

-- 
Nick Kew

Reply via email to