#escape_string_warning = off

On Mon, Aug 01, 2005 at 11:58:34AM -0700, Joshua D. Drake wrote:

What might this be?

Whether to warn on '\' in non-E'' strings.

AFAIK Bruce wants to turn this to 'on' in 8.2.

--
marko

The documentation about this is a little brief (reading from the
developer docs, section 4.1.2.1.).

Does the SQL standard provide no way to have a NULL character in a
string constant? Is single-quote the only special character?

If I have a system on 7.4 or 8.0 right now, what is the recommended
"right" way to write string constants with backslashes? I can't use E''
yet, so if I need to include a backslash it seems like there's no chance
it will be forward-compatible.

In the E'' constants, the special characters are only single-quote,
backslash, and NULL right?

Regards,
Jeff Davis

Marko Kreen wrote:

On Mon, 1 Aug 2005, Jeff Davis wrote:

Does the SQL standard provide no way to have a NULL character in a
string constant? Is single-quote the only special character?

I don't think it forbids you from using the null character. It's not like
the strings are zero terminated. Some encodings might not allow the null
character, but that's different.

ps. null character does not have anything to do with the sql NULL. I'm
sure there is someone somewhere that need this info.

--
/Dennis Bj�rklund

Dennis Bjorklund wrote:

On Mon, 1 Aug 2005, Jeff Davis wrote:

Does the SQL standard provide no way to have a NULL character in a
string constant? Is single-quote the only special character?

I don't think it forbids you from using the null character. It's not like
the strings are zero terminated. Some encodings might not allow the null
character, but that's different.

But doesn't PostgreSQL forbid us from using the NULL character in a
query at all? Don't we always have to escape or encode it in some way?
Does this new attempt at standard-compliant strings allow PostgreSQL to
accept a null character in a string?

ps. null character does not have anything to do with the sql NULL. I'm
sure there is someone somewhere that need this info.

Yeah, I was talking about '\0'.

Regards,
Jeff Davis

On Tue, 2 Aug 2005, Jeff Davis wrote:

Does the SQL standard provide no way to have a NULL character in a
string constant? Is single-quote the only special character?

I don't think it forbids you from using the null character. It's not like
the strings are zero terminated. Some encodings might not allow the null
character, but that's different.

But doesn't PostgreSQL forbid us from using the NULL character in a
query at all? Don't we always have to escape or encode it in some way?

Pg does not allow \0 in strings at all. Try SELECT 'abc\0def'; in the
current version of pg.

The sql standard doesn't forbid null values in strings as far as I know
and that's all I talked about. To have a sql standard string with null
inside you just insert the 0 byte (for normal single byte encodings), no
escaping needed.

Internally pg handles strings as \0-terminated entities which is a bit
unfortunate but that's what we have. That's why 'abc\0def' became the
string 'abc'. Most character sets forbid \0 in strings anyway.

--
/Dennis Bj�rklund

Dennis Bjorklund wrote:

On Tue, 2 Aug 2005, Jeff Davis wrote:

Does the SQL standard provide no way to have a NULL character in a
string constant? Is single-quote the only special character?

I don't think it forbids you from using the null character. It's not like
the strings are zero terminated. Some encodings might not allow the null
character, but that's different.

But doesn't PostgreSQL forbid us from using the NULL character in a
query at all? Don't we always have to escape or encode it in some way?

Pg does not allow \0 in strings at all. Try SELECT 'abc\0def'; in the
current version of pg.

The sql standard doesn't forbid null values in strings as far as I know
and that's all I talked about. To have a sql standard string with null
inside you just insert the 0 byte (for normal single byte encodings), no
escaping needed.

I guess what I'm trying to find out: does this mean that after all this
change to the way strings are handled in the future, PostgreSQL still
won't be standards-compliant for the basic '' string?

Also, let's say I have apps now in 7.4/8.0, and I want them to be
forward-compatible. Should I make a type called E so that the E''
notation will work, and then use that for strings? What is the "right"
way to do it?

I found a few things in the archives, but I didn't see these particular
things addressed.

Regards,
Jeff Davis

Am Mittwoch, 3. August 2005 01:18 schrieb Jeff Davis:

I guess what I'm trying to find out: does this mean that after all this
change to the way strings are handled in the future, PostgreSQL still
won't be standards-compliant for the basic '' string?

It will be more conforming regarding backslashes. There may be other
conformance issues, but they are not considered here.

Also, let's say I have apps now in 7.4/8.0, and I want them to be
forward-compatible. Should I make a type called E so that the E''
notation will work, and then use that for strings? What is the "right"
way to do it?

To be standards-conforming, don't use any backslash escapes. If you must use
them, use the E'' notation.

--
Peter Eisentraut
http://developer.postgresql.org/~petere/

Peter Eisentraut wrote:

Also, let's say I have apps now in 7.4/8.0, and I want them to be
forward-compatible. Should I make a type called E so that the E''
notation will work, and then use that for strings? What is the "right"
way to do it?

To be standards-conforming, don't use any backslash escapes. If you must use
them, use the E'' notation.

That doesn't really answer the question, though, since none of
7.4/8.0/8.1 interprets '' strings in a strictly standards-conforming way
as I understand it.

The impression I got from previous discussion was that you need to check
the value of the standard_compliant_strings GUC, and double backslashes
inside '' only if it was false or missing.

-O

Am Mittwoch, 3. August 2005 15:40 schrieb Oliver Jowett:

To be standards-conforming, don't use any backslash escapes. If you must
use them, use the E'' notation.

That doesn't really answer the question, though, since none of
7.4/8.0/8.1 interprets '' strings in a strictly standards-conforming way
as I understand it.

That is correct, but eventually standards_compliant_strings will be true by
default and then you have to use E'' to get backslash escapes. The above
advice is the future-proof way to go from 8.1 on.

--
Peter Eisentraut
http://developer.postgresql.org/~petere/

Am Mittwoch, 3. August 2005 15:40 schrieb Oliver Jowett:

The impression I got from previous discussion was that you need to check
the value of the standard_compliant_strings GUC, and double backslashes
inside '' only if it was false or missing.

The correct lingo would be standard_conforming_strings. I'm going to change
that.

--
Peter Eisentraut
http://developer.postgresql.org/~petere/

Peter Eisentraut wrote:

Am Mittwoch, 3. August 2005 15:40 schrieb Oliver Jowett:

The impression I got from previous discussion was that you need to check
the value of the standard_compliant_strings GUC, and double backslashes
inside '' only if it was false or missing.

The correct lingo would be standard_conforming_strings. I'm going to change
that.

Sounds good. Another question is whether this should be backpatched to
our next 7.4.X or 8.0.X release as read-only variables.

-- 
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 359-1001
  +  If your life is a hard drive,     |  13 Roberts Road
  +  Christ can be your backup.        |  Newtown Square, Pennsylvania 19073

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

Peter Eisentraut wrote:

The correct lingo would be standard_conforming_strings. I'm going to change
that.

Sounds good.

No problem here either.

Another question is whether this should be backpatched to
our next 7.4.X or 8.0.X release as read-only variables.

Unnecessary; any client code written to use this need only assume that
absence of the parameter means the old behavior.

regards, tom lane

Tom Lane wrote:

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

Peter Eisentraut wrote:

The correct lingo would be standard_conforming_strings. I'm going to change
that.

Sounds good.

No problem here either.

So does that mean for 8.1 it will be:

standard_conforming_strings = on/off

?

Joshua D. Drake wrote:

Tom Lane wrote:

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

Peter Eisentraut wrote:

The correct lingo would be standard_conforming_strings. I'm going to change
that.

Sounds good.

No problem here either.

So does that mean for 8.1 it will be:

standard_conforming_strings = on/off

off, maybe on for 8.2 or 8.3.

-- 
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 359-1001
  +  If your life is a hard drive,     |  13 Roberts Road
  +  Christ can be your backup.        |  Newtown Square, Pennsylvania 19073