On Thu, 26 Apr 2007 09:11:50 +0100, Tim Bunce <[EMAIL PROTECTED]> wrote:
> On Wed, Apr 25, 2007 at 10:40:41PM +0100, Tim Bunce wrote:
> > On Wed, Apr 25, 2007 at 01:18:08PM +0200, H.Merijn Brand wrote:
> > > > ---
> > > > Don't use DBIS or dbis when you can use DBIc_STATE (imp_xxh).
> > > > DBIS is much slower, especially on perls built with threads enabled.
> > > > It's a macro that expands to a function call, whereas
> > > > DBIc_STATE (imp_xxh) is just accessing a field in the handle structure.
> > >
> > > This does not apply to?:
> > >
> > > av = DBIS->get_fbav (imp_sth);
> > > num_fields = AvFILL (av) + 1;
> >
> > Yes. Ideally the code should have no DBIS (or dbis) in it at all. The only
> > exception would be any code that doesn't have a handle/imp_xxh handy.
>
> ... and can't be modified to have a handle/imp_xxh passed in.
I've lost you here. Is there a sane way to prevent DBIS in the tail of
the following snippet?
--8<---
AV *dbd_st_fetch (SV *sth, imp_sth_t *imp_sth)
{
int num_fields, i;
AV *av;
SV *dbh = (SV *)DBIc_PARENT_H (imp_sth);
D_imp_dbh_from_sth;
unless (use_sth_id (dbh, imp_dbh->id, imp_sth->id))
return (Nullav);
dbg (2, "DBD::Unify::st_fetch %s\n", u_sql_nm);
unless (DBIc_ACTIVE (imp_sth)) {
error (sth, -7, "fetch without open cursor");
return (Nullav);
}
/* In the next E/SQL a statement like
* "select code from table where field SHLIKE 'v_ab*'"
* will dump core in sqldfch ()
* affirmed for 6.3AB and 6.3BE
*/
EXEC SQL
FETCH :c_sql_nm
USING SQL DESCRIPTOR :o_sql_nm;
av = DBIS->get_fbav (imp_sth);
num_fields = AvFILL (av) + 1;
-->8---
I *can* change the other locations to pass a handle, but that would ruin
my code. Really. Changing all
dbg (4, " Field %3d: ", fix);
dbg (5, "[%02X %02X %02X %02X %02X] ",
(unsigned char)ftp, fln, fpr, fsc, fic);
dbg (4, "%-.8s\n", fnm);
to include handles is beyond what it would return. Maintainability is
also very important for code one doesn't visit on a daily basis.
I changed the error () function to use
DBIh_SET_ERR_CHAR (h, imp_xxh, Nullch, error_num, text, SQLSTATE, Nullch);
and behold! my DBD now supports SQLSTATE throughout :)
It's a shame that UNIFY didn't even document SQLSTATE, apart from
"The SQLSTATE variable is provided for compatibility to the
ANSI SQL2 draft standard."
So I have no idea how to extend my tests to see if the states make sense
--
H.Merijn Brand Amsterdam Perl Mongers (http://amsterdam.pm.org/)
using & porting perl 5.6.2, 5.8.x, 5.9.x on HP-UX 10.20, 11.00, 11.11,
& 11.23, SuSE 10.0 & 10.2, AIX 4.3 & 5.2, and Cygwin. http://qa.perl.org
http://mirrors.develooper.com/hpux/ http://www.test-smoke.org
http://www.goldmark.org/jeff/stupid-disclaimers/
