Re: using array of char pointers gives wrong results

Hi,
When array of char * is used as target for the FETCH statement returning
more than one row, it tries to store all the result in the first element.
PFA test_char_select.pgc, which fetches first 3 relnames from pg_class
ordered by relname. The program prints following result

steps to compile and build the program
ecpg -c -I<ecpg_include_dir> test_char_select.pgc
cc -I<pg installation include dir> -g -c -o test_char_select.o
test_char_select.c
cc -g test_char_select.o -L<pg installation lib dir> -lecpg -lpq
-lpgtypes -o test_char_select

output
./test_char_select
relname=___pg_foreign_table_columns
relname=
relname=

The first three relnames should have been
postgres=# select relname from pg_class order by relname limit 3;
relname
---------------------------
_pg_foreign_data_wrappers
_pg_foreign_servers
_pg_foreign_table_columns

It's obvious that the first element of the array is being overwritten with
an offset of 1.

This happens because, the array of char pointer is dumped as
/* Fetch multiple columns into one structure. */
{ ECPGdo(__LINE__, 0, 1, NULL, 0, ECPGst_normal, "fetch 3 from cur1",
ECPGt_EOIT,
ECPGt_char,(strings),(long)0,(long)3,*(1)*sizeof(char)*,
ECPGt_NO_INDICATOR, NULL , 0L, 0L, 0L, ECPGt_EORT);

Since the offset is 1, the next result overwrites the previous result
except for the first byte.

PFA patch ecpg_char_ptr_arr.patch to fix this issue. It has changes as
follows
1. Dump array of char pointer with right offset i.e. sizeof(char *)
2. While reading array of char pointer in ecpg_do_prologue(), use the
address instead of the value at that address
3. The pointer arithmetic should treat such variable as char **, instead of
char *

ECPG regression tests do not show any failures with this patch.
--
Best Wishes,
Ashutosh Bapat
EnterpriseDB Corporation
The Postgres Database Company

Let me bring the bug fix again to the surface. Is anybody looking at this
fix?

On Tue, Apr 29, 2014 at 2:25 PM, Ashutosh Bapat <
ashutosh(dot)bapat(at)enterprisedb(dot)com> wrote:

> Hi,
> When array of char * is used as target for the FETCH statement returning
> more than one row, it tries to store all the result in the first element.
> PFA test_char_select.pgc, which fetches first 3 relnames from pg_class
> ordered by relname. The program prints following result
>
> steps to compile and build the program
> ecpg -c -I<ecpg_include_dir> test_char_select.pgc
> cc -I<pg installation include dir> -g -c -o test_char_select.o
> test_char_select.c
> cc -g test_char_select.o -L<pg installation lib dir> -lecpg -lpq
> -lpgtypes -o test_char_select
>
> output
> ./test_char_select
> relname=___pg_foreign_table_columns
> relname=
> relname=
>
> The first three relnames should have been
> postgres=# select relname from pg_class order by relname limit 3;
> relname
> ---------------------------
> _pg_foreign_data_wrappers
> _pg_foreign_servers
> _pg_foreign_table_columns
>
> It's obvious that the first element of the array is being overwritten with
> an offset of 1.
>
> This happens because, the array of char pointer is dumped as
> /* Fetch multiple columns into one structure. */
> { ECPGdo(__LINE__, 0, 1, NULL, 0, ECPGst_normal, "fetch 3 from cur1",
> ECPGt_EOIT,
> ECPGt_char,(strings),(long)0,(long)3,*(1)*sizeof(char)*,
> ECPGt_NO_INDICATOR, NULL , 0L, 0L, 0L, ECPGt_EORT);
>
> Since the offset is 1, the next result overwrites the previous result
> except for the first byte.
>
> PFA patch ecpg_char_ptr_arr.patch to fix this issue. It has changes as
> follows
> 1. Dump array of char pointer with right offset i.e. sizeof(char *)
> 2. While reading array of char pointer in ecpg_do_prologue(), use the
> address instead of the value at that address
> 3. The pointer arithmetic should treat such variable as char **, instead
> of char *
>
> ECPG regression tests do not show any failures with this patch.
> --
> Best Wishes,
> Ashutosh Bapat
> EnterpriseDB Corporation
> The Postgres Database Company
>

--
Best Wishes,
Ashutosh Bapat
EnterpriseDB Corporation
The Postgres Database Company

> PFA patch ecpg_char_ptr_arr.patch to fix this issue. It has changes as
> follows
> ...

Thanks for finding and fixing the problem. Patch applied to complete 9 series with only a minor change.

Michael
--
Michael Meskes
Michael at Fam-Meskes dot De, Michael at Meskes dot (De|Com|Net|Org)
Michael at BorussiaFan dot De, Meskes at (Debian|Postgresql) dot Org
Jabber: michael.meskes at gmail dot com
VfL Borussia! Força Barça! Go SF 49ers! Use Debian GNU/Linux, PostgreSQL

Thanks Michael.

On Tue, May 6, 2014 at 5:01 PM, Michael Meskes <meskes(at)postgresql(dot)org>wrote:

> > PFA patch ecpg_char_ptr_arr.patch to fix this issue. It has changes as
> > follows
> > ...
>
> Thanks for finding and fixing the problem. Patch applied to complete 9
> series with only a minor change.
>
> Michael
> --
> Michael Meskes
> Michael at Fam-Meskes dot De, Michael at Meskes dot (De|Com|Net|Org)
> Michael at BorussiaFan dot De, Meskes at (Debian|Postgresql) dot Org
> Jabber: michael.meskes at gmail dot com
> VfL Borussia! Força Barça! Go SF 49ers! Use Debian GNU/Linux, PostgreSQL
>

--
Best Wishes,
Ashutosh Bapat
EnterpriseDB Corporation
The Postgres Database Company