Re: Patch for undefined value error when zero bind values
On Mon, Jan 26, 2004 at 03:26:32PM -, Avis, Ed wrote: > [diff to DBI.pm] > > >+ ($tuple_status) ? @$tuple_status = () : $tuple_status = []; > > AFAIK this code is equivalent to > > @$tuple_status = (); > > because of autovivification. True. I prefer to make it explicit. I guess a comment that it autovivifies would have sufficed. Ho hum. Tim.
RE: Patch for undefined value error when zero bind values
[diff to DBI.pm] >+ ($tuple_status) ? @$tuple_status = () : $tuple_status = []; AFAIK this code is equivalent to @$tuple_status = (); because of autovivification. But anyway, it works. -- Ed Avis <[EMAIL PROTECTED]>
Re: Patch for undefined value error when zero bind values
On Mon, Jan 26, 2004 at 02:21:33PM -, Avis, Ed wrote:
> Tim Bunce <[EMAIL PROTECTED]> wrote:
>
> >Actualy I did this instead, as it seems more correct:
> >
> >@@ -1720,5 +1721,6 @@
> > sub execute_for_fetch {
> >my ($sth, $fetch_tuple_sub, $tuple_status) = @_;
> >- @$tuple_status = () if $tuple_status; # reset the status array
> >+ # start with empty status array
> >+ ($tuple_status) ? @$tuple_status = () : $tuple_status = undef;
>
> But when I try that code it fails the tests I added.
Uh, groan. The diff I pasted was an old one where I was rechecking
that the tests failed by replacing [] with undef. This is the
right diff hunk:
@@ -1720,5 +1721,6 @@
sub execute_for_fetch {
my ($sth, $fetch_tuple_sub, $tuple_status) = @_;
- @$tuple_status = () if $tuple_status; # reset the status array
+ # start with empty status array
+ ($tuple_status) ? @$tuple_status = () : $tuple_status = [];
my ($err_count, %errstr_cache);
> Did you apply the patch to the test suite?
Yes, thanks. (I just had too many windows with diff in them :)
Tim.
RE: Patch for undefined value error when zero bind values
Tim Bunce <[EMAIL PROTECTED]> wrote:
>Actualy I did this instead, as it seems more correct:
>
>@@ -1720,5 +1721,6 @@
> sub execute_for_fetch {
>my ($sth, $fetch_tuple_sub, $tuple_status) = @_;
>- @$tuple_status = () if $tuple_status; # reset the status array
>+ # start with empty status array
>+ ($tuple_status) ? @$tuple_status = () : $tuple_status = undef;
But when I try that code it fails the tests I added. Did you apply
the patch to the test suite?
--
Ed Avis <[EMAIL PROTECTED]>
Re: Patch for undefined value error when zero bind values
On Mon, Jan 26, 2004 at 01:18:19PM +, Tim Bunce wrote:
> Thanks, applied.
>
> Tim.
>
> On Mon, Jan 26, 2004 at 10:18:03AM -, Avis, Ed wrote:
> > This patch to DBI-1.40 fixes an error
> >
> > Can't use an undefined value as an ARRAY reference at DBI.pm line 1735
> >
> > when using execute_array() with zero values for a bind parameter. It
> > includes some new test cases, but not all of the new tests failed
> > before.
> >
> > diff -ru DBI-1.40/DBI.pm DBI-1.40-new/DBI.pm
> > --- DBI-1.40/DBI.pm 2004-01-08 14:03:57.0 +
> > +++ DBI-1.40-new/DBI.pm 2004-01-26 10:15:46.0 +
> > @@ -1719,7 +1719,7 @@
> >
> > sub execute_for_fetch {
> > my ($sth, $fetch_tuple_sub, $tuple_status) = @_;
> > - @$tuple_status = () if $tuple_status; # reset the status array
> > + @$tuple_status = (); # reset the status array
Actualy I did this instead, as it seems more correct:
@@ -1720,5 +1721,6 @@
sub execute_for_fetch {
my ($sth, $fetch_tuple_sub, $tuple_status) = @_;
- @$tuple_status = () if $tuple_status; # reset the status array
+ # start with empty status array
+ ($tuple_status) ? @$tuple_status = () : $tuple_status = undef;
I also spotted that the docs don't match the code so I changed the docs:
-The number of tuples executed is returned, regardless of the success
-or failure of those executions. Use tuple_status to check.
+The number of tuples executed is returned I there were no errors.
+If there were any errors then C is returned and the @tuple_status
+array can be used to discover which tuples failed and with what errors.
Thanks.
Tim.
Re: Patch for undefined value error when zero bind values
Thanks, applied.
Tim.
On Mon, Jan 26, 2004 at 10:18:03AM -, Avis, Ed wrote:
> This patch to DBI-1.40 fixes an error
>
> Can't use an undefined value as an ARRAY reference at DBI.pm line 1735
>
> when using execute_array() with zero values for a bind parameter. It
> includes some new test cases, but not all of the new tests failed
> before.
>
> diff -ru DBI-1.40/DBI.pm DBI-1.40-new/DBI.pm
> --- DBI-1.40/DBI.pm 2004-01-08 14:03:57.0 +
> +++ DBI-1.40-new/DBI.pm 2004-01-26 10:15:46.0 +
> @@ -1719,7 +1719,7 @@
>
> sub execute_for_fetch {
> my ($sth, $fetch_tuple_sub, $tuple_status) = @_;
> - @$tuple_status = () if $tuple_status; # reset the status array
> + @$tuple_status = (); # reset the status array
>
> my ($err_count, %errstr_cache);
> while ( my $tuple = &$fetch_tuple_sub() ) {
> diff -ru DBI-1.40/t/15array.t DBI-1.40-new/t/15array.t
> --- DBI-1.40/t/15array.t 2003-08-20 01:15:34.0 +0100
> +++ DBI-1.40-new/t/15array.t 2004-01-26 10:15:15.0 +
> @@ -3,7 +3,7 @@
> use strict;
> use Test;
>
> -BEGIN { plan tests => 34 }
> +BEGIN { plan tests => 39 }
>
> use Data::Dumper;
> $Data::Dumper::Indent = 0;
> @@ -68,6 +68,25 @@
> $dumped = Dumper($tuple_status);
> ok( $dumped, "[1,1,1]");
>
> +# --- with no values for bind params, should execute zero times
> +
> [EMAIL PROTECTED] = ();
> +ok( $sth->execute_array( { ArrayTupleStatus => $tuple_status },
> +[], [], [], [],
> +),
> +0);
> +ok( @$rows, 0 );
> +ok( @$tuple_status, 0 );
> +
> +# --- catch 'undefined value' bug with zero bind values
> +
> [EMAIL PROTECTED] = ();
> +my $sth_other = $dbh->prepare("insert", {
> + rows => $rows, # where to 'insert' (push) the rows
> + NUM_OF_PARAMS => 1,
> +});
> +ok( $sth_other->execute_array( {}, [] ), 0 );
> +ok( @$rows, 0);
>
> # --- ArrayTupleFetch code-ref tests ---
>
>
> --
> Ed Avis <[EMAIL PROTECTED]>
