Bug in gram.y?

It appears to me that there is a bug in gram.y. From looking at the code
(while working on ecpg) it seems that it doesn't accept a createdb with th
eoption: with location = ... but without the option with encoding = ...

Looks fine to me:

opt_database1: LOCATION '=' location { $$ = $3; }
| /*EMPTY*/ { $$ = NULL; }
;

opt_database2: ENCODING '=' encoding { $$ = $3; }
| /*EMPTY*/ { $$ = NULL; }
;

-- 
Bruce Momjian                          |  830 Blythe Avenue
maillist@candle.pha.pa.us              |  Drexel Hill, Pennsylvania 19026
  +  If your life is a hard drive,     |  (610) 353-9879(w)
  +  Christ can be your backup.        |  (610) 853-3000(h)

it seems that it doesn't accept a createdb with the
option location = ... but without the option encoding = ...

I think the code is sub-optimal for a couple of reasons:

1) LOCATION must be the first argument, ENCODING must be the second. The
number and order of any options should be flexible.

2) the options should have a comma delimiter between them. Currently
they do not. To introduce the comma delimiter will require restructuring
that part of the parsing, but isn't a big deal.

I'm pretty sure that the options can be specified individually, but it
does not generalize well. The two specific problems above should be
considered fatal for a final version :/

I've worked through problems like this for other SQL92 syntax, and can
probably fix up this one too. Will put it on my list unless someone else
wants it...

- Tom

it seems that it doesn't accept a createdb with the
option location = ... but without the option encoding = ...

I think the code is sub-optimal for a couple of reasons:

1) LOCATION must be the first argument, ENCODING must be the second. The
number and order of any options should be flexible.

2) the options should have a comma delimiter between them. Currently
they do not. To introduce the comma delimiter will require restructuring
that part of the parsing, but isn't a big deal.

I'm pretty sure that the options can be specified individually, but it
does not generalize well. The two specific problems above should be
considered fatal for a final version :/

I've worked through problems like this for other SQL92 syntax, and can
probably fix up this one too. Will put it on my list unless someone else
wants it...

I'm responsible for that problem. But seems you have a better
solution. Please take it.
--
Tatsuo Ishii
t-ishii@sra.co.jp

On Mon, Aug 03, 1998 at 02:19:44PM -0400, Bruce Momjian wrote:

Looks fine to me:

opt_database1: LOCATION '=' location { $$ = $3; }
| /*EMPTY*/ { $$ = NULL; }
;

opt_database2: ENCODING '=' encoding { $$ = $3; }
| /*EMPTY*/ { $$ = NULL; }
;

Yes, but check the code under CreatedbStmt:

n->dbpath = $5;
#ifdef MULTIBYTE
if ($6 != NULL) {
n->encoding = pg_char_to_encoding($6);
if (n->encoding < 0) {
elog(ERROR, "invalid encoding name %s", $6);
}
} else {
n->encoding = GetTemplateEncoding();
}
#else
elog(ERROR, "WITH ENCODING is not supported");
#endif
$$ = (Node *)n;

If you have undefined MULTIBYTE you will get the WITH ENCODING error message
despite the command being legal even without this option.

Michael
--
Dr. Michael Meskes meskes@online-club.de, meskes@debian.org
Go SF49ers! Go Rhein Fire! Use Debian GNU/Linux!

On Mon, Aug 03, 1998 at 02:19:44PM -0400, Bruce Momjian wrote:

Looks fine to me:

opt_database1: LOCATION '=' location { $$ = $3; }
| /*EMPTY*/ { $$ = NULL; }
;

opt_database2: ENCODING '=' encoding { $$ = $3; }
| /*EMPTY*/ { $$ = NULL; }
;

Yes, but check the code under CreatedbStmt:

n->dbpath = $5;
#ifdef MULTIBYTE
if ($6 != NULL) {
n->encoding = pg_char_to_encoding($6);
if (n->encoding < 0) {
elog(ERROR, "invalid encoding name %s", $6);
}
} else {
n->encoding = GetTemplateEncoding();
}
#else
elog(ERROR, "WITH ENCODING is not supported");
#endif
$$ = (Node *)n;

If you have undefined MULTIBYTE you will get the WITH ENCODING error message
despite the command being legal even without this option.

Good point. I have changed it to:

} else {
n->encoding = GetTemplateEncoding();
}
#else
if ($6 != NULL)

^^^^^^^^^^^^^^^

elog(ERROR, "WITH ENCODING is not supported");
#endif
$$ = (Node *)n;

-- 
Bruce Momjian                          |  830 Blythe Avenue
maillist@candle.pha.pa.us              |  Drexel Hill, Pennsylvania 19026
  +  If your life is a hard drive,     |  (610) 353-9879(w)
  +  Christ can be your backup.        |  (610) 853-3000(h)