On Fri, 2006-11-10 at 17:24 -0500, Tom Lane wrote:
> Theo Kramer <theo@flame.co.za> writes:
> > On Fri, 2006-11-10 at 10:50 +0100, Peter Eisentraut wrote:
> >> The patch is backwards.
>
> > Thanks - re-done and re-attached
>
> I applied the part of this that reformats the PQexecParams argument
> descriptions --- I agree that makes it more readable. I did not like
> the proposed table of datatypes though. It strikes me as incomplete,
> unhelpful, and in places downright misleading. For instance, I don't
> see the point of listing ODBC type codes; they are irrelevant to libpq,
> and the only likely outcome of listing them is that someone might think
> he could use them for paramTypes[] entries. The description of the
> binary formats was not helpful either, as it avoided any of the useful
> details (endianness etc), as well as portability issues like whether
> timestamps are int or float.
>
Thanks for the input - I agree that the table is incomplete.
My intention with the table is for it to be a start to something that is
more useful than the incomplete examples regarding these functions. I
suggest I redo it taking into consideration your input, unless you have
other suggestions for improving the docs on the use with typical values
for the arguments.
On the ODBC types - I found these useful as my interface is based on
that. I imagine that others who have used ODBC and who would like to use
the libpq functions may also find it useful, but also agree that it may
be misleading.
Perhaps, with a couple of iterations, we can get to something that will
make the documentation on these functions more useful.
--
Regards
Theo
