CURRENT/OLD keywords still broken

peter=# select 1 as current;
old
-----
1

This is now the inverse of what it used to do, but it's still not what it
*should* do. I see someone already tried to fix that (keywords.c 1.76 ->
1.77, TODO list), but he should try again.

That was me. The old code did old -> current, so I changed it to do
current -> old. How else can I fix this? Attached is the old patch.

-- 
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026

Peter Eisentraut <peter_e@gmx.net> writes:

peter=# select 1 as current;
old
-----
1

This is now the inverse of what it used to do, but it's still not what it
*should* do. I see someone already tried to fix that (keywords.c 1.76 ->
1.77, TODO list), but he should try again.

We should rip out the whole current/old aliasing, IMHO. CURRENT has
been unsupported for a version or two now, hasn't it?

I had that on my personal TODO list, but I was going to leave it till
after 7.1 because I had mistakenly thought that CURRENT was still a
keyword in 7.0. But it wasn't, was it? Bruce, why did you put in
that mapping?

regards, tom lane

Bruce Momjian <pgman@candle.pha.pa.us> writes:

That was me. The old code did old -> current, so I changed it to do
current -> old. How else can I fix this? Attached is the old patch.

But CURRENT was strictly an internal token name, not a string the user
could actually see.  So there's no need to have
+ 	/* for portability with old rules bjm 2000-06-12 */
+ 	{"current", OLD},

The only way that there would be a compatibility problem would be if
ruleutils.c had been set up to print CURRENT, but it wasn't:

*** 1278,1284 ****
                                         quote_identifier(rte->relname));
                    else if (!strcmp(rte->ref->relname, "*NEW*"))
                        appendStringInfo(buf, "new.");
!                       else if (!strcmp(rte->ref->relname, "*CURRENT*"))
                        appendStringInfo(buf, "old.");
                    else
                        appendStringInfo(buf, "%s.",
--- 1278,1284 ----
                                         quote_identifier(rte->relname));
                    else if (!strcmp(rte->ref->relname, "*NEW*"))
                        appendStringInfo(buf, "new.");
!                       else if (!strcmp(rte->ref->relname, "*OLD*"))
                        appendStringInfo(buf, "old.");
                    else

NEW and OLD are what the user see, and have been for awhile. So there's
no compatibility issue here.

regards, tom lane

OK, compatibility mapping removed.

Bruce Momjian <pgman@candle.pha.pa.us> writes:

That was me. The old code did old -> current, so I changed it to do
current -> old. How else can I fix this? Attached is the old patch.

But CURRENT was strictly an internal token name, not a string the user
could actually see.  So there's no need to have
+ 	/* for portability with old rules bjm 2000-06-12 */
+ 	{"current", OLD},

The only way that there would be a compatibility problem would be if
ruleutils.c had been set up to print CURRENT, but it wasn't:

*** 1278,1284 ****
quote_identifier(rte->relname));
else if (!strcmp(rte->ref->relname, "*NEW*"))
appendStringInfo(buf, "new.");
!                       else if (!strcmp(rte->ref->relname, "*CURRENT*"))
appendStringInfo(buf, "old.");
else
appendStringInfo(buf, "%s.",
--- 1278,1284 ----
quote_identifier(rte->relname));
else if (!strcmp(rte->ref->relname, "*NEW*"))
appendStringInfo(buf, "new.");
!                       else if (!strcmp(rte->ref->relname, "*OLD*"))
appendStringInfo(buf, "old.");
else

NEW and OLD are what the user see, and have been for awhile. So there's
no compatibility issue here.

regards, tom lane

-- 
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026

Bruce Momjian writes:

peter=# select 1 as current;
old
-----
1

This is now the inverse of what it used to do, but it's still not what it
*should* do. I see someone already tried to fix that (keywords.c 1.76 ->
1.77, TODO list), but he should try again.

That was me. The old code did old -> current, so I changed it to do
current -> old. How else can I fix this? Attached is the old patch.

Maybe add another branch into the SpecialRuleRelation grammar rule with
CURRENT instead of OLD.

--
Peter Eisentraut peter_e@gmx.net http://yi.org/peter-e/