ERROR: operator does not exist: integer !=- integer

Here is the steps to reproduce it in CVS HEAD:

$ uname -a
Linux os-server 2.6.9-11.19AX #1 Fri Aug 5 05:12:07 EDT 2005 i686 i686 i386
GNU/Linux

$ ./postgres --single -D $HOME/pgsql/data postgres

PostgreSQL stand-alone backend 8.3devel
backend> show server_version;
1: server_version (typeid = 25, len = -1, typmod = -1, byval =
f)
----
1: server_version = "8.3devel" (typeid = 25, len = -1, typmod = -1,
byval = f)
----
backend> select -1 != -1;
1: ?column? (typeid = 16, len = 1, typmod = -1, byval = t)
----
1: ?column? = "f" (typeid = 16, len = 1, typmod = -1, byval =
t)
----
backend> select -1 !=-1;
ERROR: operator does not exist: integer !=- integer at character 11
HINT: No operator matches the given name and argument type(s). You might
need to add explicit type casts.
STATEMENT: select -1 !=-1;

A quick hack in scan.l :

*** src/backend/parser/scan.l.old 2007-03-04 11:39:56.831289992 +0800
--- src/backend/parser/scan.l 2007-03-04 11:40:04.142178568 +0800
***************
*** 605,610 ****
--- 605,617 ----
{
int ic;

+ /* filter out operaters end
with '=' */
+ if (yytext[nchars - 2] ==
'=')
+ {
+ nchars--;
+ continue;
+ }
+
for (ic = nchars-2; ic >= 0;
ic--)
{
if
(strchr("~!(at)#^&|`?%", yytext[ic]))

Now the result is correct:

backend> select -1 !=-1;
1: ?column? (typeid = 16, len = 1, typmod = -1, byval = t)
----
1: ?column? = "f" (typeid = 16, len = 1, typmod = -1, byval =
t)
----
--
Regards,
William ZHANG

On 2007-03-04, William ZHANG <uniware(at)zedware(dot)org> wrote:
> Here is the steps to reproduce it in CVS HEAD:
> backend> select -1 !=-1;

This arguably isn't a bug, because != is not a standard SQL operator, and
therefore !=- can legitimately be defined as a single operator by the user.

--
Andrew, Supernews
http://www.supernews.com - individual and corporate NNTP services

Andrew - Supernews wrote:
> On 2007-03-04, William ZHANG <uniware(at)zedware(dot)org> wrote:
>> Here is the steps to reproduce it in CVS HEAD:
>> backend> select -1 !=-1;
>
> This arguably isn't a bug, because != is not a standard SQL operator, and
> therefore !=- can legitimately be defined as a single operator by the user.
>

I missed the first post and can't seem to search for it - so correct me
if I am missing something.

Isn't the problem here a missing space? != is a valid operator and -1 is
the value you are comparing to. !=-1 is not valid but != -1 is correct
and what I assume you are looking to achieve.

The negation operator goes with the int being negated and is not part of
the comparison operator != the space is needed there to separate the two.

--

Shane Ambler
pgSQL(at)Sheeky(dot)Biz

Get Sheeky @ http://Sheeky.Biz

"William ZHANG" <uniware(at)zedware(dot)org> writes:
> backend> select -1 !=-1;
> ERROR: operator does not exist: integer !=- integer at character 11

This is not a bug.

regards, tom lane

>
> I missed the first post and can't seem to search for it - so correct
> me if I am missing something.
>
> Isn't the problem here a missing space? != is a valid operator and -1
> is the value you are comparing to. !=-1 is not valid but != -1 is
> correct and what I assume you are looking to achieve.
>

Well yes it will work if you add a space, but technically the problem is
the query should be written like this:

1 <>-1 or 1 <> -1

Joshua D. Drake

> The negation operator goes with the int being negated and is not part
> of the comparison operator != the space is needed there to separate
> the two.
>
>
>

I get it. scan.l converts '!=' to '<>':

644 /* Convert "!=" operator to "<>" for
compatibility */
645 if (strcmp(yytext, "!=") == 0)
646 yylval.str = pstrdup("<>");
647 else
648 yylval.str = pstrdup(yytext);

""Joshua D. Drake"" <jd(at)commandprompt(dot)com>
>
> Well yes it will work if you add a space, but technically the problem is
> the query should be written like this:
>
> 1 <>-1 or 1 <> -1
>
> Joshua D. Drake