| From: | Abhijit Menon-Sen <ams(at)oryx(dot)com> |
|---|---|
| To: | Gevik Babakhani <pgdev(at)xs4all(dot)nl> |
| Cc: | pgsql-hackers(at)postgresql(dot)org |
| Subject: | Re: help wanted (fmgr.c) |
| Date: | 2006-09-11 08:53:41 |
| Message-ID: | 20060911085341.GC31654@penne.toroid.org |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
At 2006-09-11 10:25:22 +0200, pgdev(at)xs4all(dot)nl wrote:
>
> What are we testing to be NULL here?
> Do we expect str to changed at line 1715
No. (Read the comment just above the function.)
The code is like this, starting from line 1703:
if (str == NULL && flinfo->fn_strict)
return (Datum) 0;
That is, if the function is declared strict, and the argument (str) is
0, just return NULL straightaway. Then it sets up the fcinfo and calls
the function, and then:
...
/* Should get null result if and only if str is NULL */
if (str == NULL)
{
if (!fcinfo.isnull)
elog(ERROR, "input function %u returned non-NULL",
fcinfo.flinfo->fn_oid);
}
else
{
if (fcinfo.isnull)
elog(ERROR, "input function %u returned NULL",
fcinfo.flinfo->fn_oid);
}
This says: If the argument is NULL and the input function didn't return
a NULL, log an error; but if the argument is non-NULL and the function
returned NULL, log this other error. (Note that a function would set
fcinfo->isnull to indicate that it wants to return an SQL NULL, as
explained in $DOC/plhandler.html)
-- ams
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Adrian Maier | 2006-09-11 09:14:33 | Cassowary failing to report the results back to the farm |
| Previous Message | Reinhard Max | 2006-09-11 08:49:54 | Re: [Pgsqlrpms-hackers] Safer auto-initdb for RPM init |