Обсуждение: 0/1 vs true/false
Just having a small argument with an application developer ... is using 0/1 for boolean types SQL compliant? I am trying to convince him that the proper SQL compliant (and postgresql compliant) syntax is true/false but he won't budge ... The app as currently written no longer works with postgres because they code uses 0/1 instead of the now enforced true/false for boolean types. Can someone point me to an SQL spec and section where this is clearly stated out? Thanks, Jean-Christian Imbeault
I ran into this problem some time ago. I ended up using a query with a CASE statement in it to caste the boolean as a 1 or 0. I guess one could do this as a view as well. Jean-Christian Imbeault wrote: > Just having a small argument with an application developer ... > > is using 0/1 for boolean types SQL compliant? I am trying to convince > him that the proper SQL compliant (and postgresql compliant) syntax is > true/false but he won't budge ... > > The app as currently written no longer works with postgres because > they code uses 0/1 instead of the now enforced true/false for boolean > types. > > Can someone point me to an SQL spec and section where this is clearly > stated out? > > Thanks, > > Jean-Christian Imbeault > > > ---------------------------(end of broadcast)--------------------------- > TIP 3: if posting/reading through Usenet, please send an appropriate > subscribe-nomail command to majordomo@postgresql.org so that your > message can get through to the mailing list cleanly > >
Jean-Christian Imbeault writes: > is using 0/1 for boolean types SQL compliant? No. > Can someone point me to an SQL spec and section where this is clearly > stated out? Since it's not compliant, you won't find it anywhere in the standard. -- Peter Eisentraut peter_e@gmx.net
Peter Eisentraut wrote: > > Since it's not compliant, you won't find it anywhere in the standard. Oops, I meant to say can someone point me to a ressource (SQL standard and section?) which states that true/false must be used for booleans :) Jean-Christian Imbeault
I had do do this sort of thing for a some developers. It was actually 'true' and 'false' that was wanted not the 0/1. I wrote a little plgsql function and installed it as a cast to text. -- Nigel J. Andrews On Wed, 23 Jul 2003, Chris Travers wrote: > I ran into this problem some time ago. I ended up using a query with a > CASE statement in it to caste the boolean as a 1 or 0. I guess one could > do this as a view as well. > > Jean-Christian Imbeault wrote: > > > Just having a small argument with an application developer ... > > > > is using 0/1 for boolean types SQL compliant? I am trying to convince > > him that the proper SQL compliant (and postgresql compliant) syntax is > > true/false but he won't budge ... > > > > The app as currently written no longer works with postgres because > > they code uses 0/1 instead of the now enforced true/false for boolean > > types. > > > > Can someone point me to an SQL spec and section where this is clearly > > stated out? > > > > Thanks, > > > > Jean-Christian Imbeault
Nigel J. Andrews wrote: > > I had do do this sort of thing for a some developers. It was actually 'true' > and 'false' that was wanted not the 0/1. I wrote a little plgsql function and > installed it as a cast to text. Thanks. I know that there are workarounds, I just want to "prove" that 0/1 is not standards compliant. If I can do that I am pretty sure that the developer will have nothing against changing to true/false. Right not he just things postgresql must be broken because 0/1 works in MySQL ... But he's pretty good when it comes to fixing things because they don't follow standards. So if I can show him his SQL isn't standards compliant he'll probably happily fix it. Jean-Christian Imbeault
Jean-Christian Imbeault wrote:
> Just having a small argument with an application developer ...
>
> is using 0/1 for boolean types SQL compliant? I am trying to convince
> him that the proper SQL compliant (and postgresql compliant) syntax is
> true/false but he won't budge ...
>
> The app as currently written no longer works with postgres because they
> code uses 0/1 instead of the now enforced true/false for boolean types.
>
> Can someone point me to an SQL spec and section where this is clearly
> stated out?
Would this be what you are looking for:
ISO/IEC 9075-2:1999 (E) ©ISO/IEC
5.3 <literal>
(..)
<boolean literal> ::=
TRUE
| FALSE
| UNKNOWN
Additionally about UNKNOWN:
ISO/IEC 9075-2:1999 (E) ©ISO/IEC
4.6 Boolean types
The data type boolean comprises the distinct truth values true
and false . Unless prohibited by a NOT NULL constraint, the
boolean data type also supports the unknown truth value as the
null value. This specification does not make a distinction
between the null value of the boolean data type and the unknown
truth value that is the result of an SQL <predicate>, <search
condition>, or <boolean value expression>; they may be used
interchangeably to mean exactly the same thing.
HTH,
Jochem
This makes me wonder, what about 't' and 'f'?... will they disappear in newer versions of postgreSQL?
On Wed, 2003-07-23 at 09:21, Jochem van Dieten wrote:
On Wed, 2003-07-23 at 09:21, Jochem van Dieten wrote:
Jean-Christian Imbeault wrote:
> Just having a small argument with an application developer ...
>
> is using 0/1 for boolean types SQL compliant? I am trying to convince
> him that the proper SQL compliant (and postgresql compliant) syntax is
> true/false but he won't budge ...
>
> The app as currently written no longer works with postgres because they
> code uses 0/1 instead of the now enforced true/false for boolean types.
>
> Can someone point me to an SQL spec and section where this is clearly
> stated out?
Would this be what you are looking for:
ISO/IEC 9075-2:1999 (E) ©ISO/IEC
5.3 <literal>
(..)
<boolean literal> ::= TRUE | FALSE | UNKNOWN
Additionally about UNKNOWN:
ISO/IEC 9075-2:1999 (E) ©ISO/IEC
4.6 Boolean types
The data type boolean comprises the distinct truth values true
and false . Unless prohibited by a NOT NULL constraint, the
boolean data type also supports the unknown truth value as the
null value. This specification does not make a distinction
between the null value of the boolean data type and the unknown
truth value that is the result of an SQL <predicate>, <search
condition>, or <boolean value expression>; they may be used
interchangeably to mean exactly the same thing.
HTH,
Jochem
---------------------------(end of broadcast)---------------------------
TIP 5: Have you checked our extensive FAQ?
http://www.postgresql.org/docs/faqs/FAQ.html
Вложения
> Can someone point me to an SQL spec and section where this is clearly > stated out? > You may want to review the files mentioned at http://archives.postgresql.org/pgsql-sql/2000-04/msg00118.php hth cl.
Jean-Christian Imbeault <jc@mega-bucks.co.jp> writes:
> Oops, I meant to say can someone point me to a ressource (SQL standard
> and section?) which states that true/false must be used for booleans :)
Well, there is no boolean type per se in SQL92. But there is in SQL99.
I think the most relevant part of the spec is the definition of boolean
literals in section 5.3:
<boolean literal> ::=
TRUE
| FALSE
| UNKNOWN
Note that the spec only really speaks to the question of what booleans
look like when written as constants in SQL statements. AFAICT it does
not take a position on what representations are to be used when
transferring data into or out of the database. So using '0' or '1' as
input to a boolean field is not contrary to spec, but merely outside
the spec (and in fact we will accept those strings for boolean...)
regards, tom lane
I doubt it, same way '0' and '1' have not disappeared despite 0 and 1 have. There is still a function that can convert a string to a boolean. There is no function to convert an integer to a boolean, which is why 0 and 1 do not work but '0', 'f', '1', 't' all work.
Terry Fielder
Manager Software Development and Deployment
Great Gulf Homes / Ashton Woods Homes
terry@greatgulfhomes.com
Fax: (416) 441-9085
-----Original Message-----This makes me wonder, what about 't' and 'f'?... will they disappear in newer versions of postgreSQL?
From: pgsql-general-owner@postgresql.org [mailto:pgsql-general-owner@postgresql.org]On Behalf Of Franco Bruno Borghesi
Sent: Wednesday, July 23, 2003 10:21 AM
To: Jochem van Dieten
Cc: Jean-Christian Imbeault; pgsql-general@postgresql.org
Subject: Re: [GENERAL] 0/1 vs true/false
On Wed, 2003-07-23 at 09:21, Jochem van Dieten wrote:Jean-Christian Imbeault wrote: > Just having a small argument with an application developer ... > > is using 0/1 for boolean types SQL compliant? I am trying to convince > him that the proper SQL compliant (and postgresql compliant) syntax is > true/false but he won't budge ... > > The app as currently written no longer works with postgres because they > code uses 0/1 instead of the now enforced true/false for boolean types. > > Can someone point me to an SQL spec and section where this is clearly > stated out? Would this be what you are looking for: ISO/IEC 9075-2:1999 (E) ©ISO/IEC 5.3 <literal> (..) <boolean literal> ::= TRUE | FALSE | UNKNOWN Additionally about UNKNOWN: ISO/IEC 9075-2:1999 (E) ©ISO/IEC 4.6 Boolean types The data type boolean comprises the distinct truth values true and false . Unless prohibited by a NOT NULL constraint, the boolean data type also supports the unknown truth value as the null value. This specification does not make a distinction between the null value of the boolean data type and the unknown truth value that is the result of an SQL <predicate>, <search condition>, or <boolean value expression>; they may be used interchangeably to mean exactly the same thing. HTH, Jochem ---------------------------(end of broadcast)--------------------------- TIP 5: Have you checked our extensive FAQ? http://www.postgresql.org/docs/faqs/FAQ.html
Franco Bruno Borghesi <franco@akyasociados.com.ar> writes:
> This makes me wonder, what about 't' and 'f'?... will they disappear in
> newer versions of postgreSQL?
No. See my comment about I/O representations being outside the spec.
There is no inconsistency in the following examples:
regression=# select true;
bool
------
t
(1 row)
regression=# select t;
ERROR: attribute "t" not found
regression=# select 't'::boolean;
bool
------
t
(1 row)
regression=# select '1'::boolean;
bool
------
t
(1 row)
regards, tom lane
Tom Lane writes: > Note that the spec only really speaks to the question of what booleans > look like when written as constants in SQL statements. AFAICT it does > not take a position on what representations are to be used when > transferring data into or out of the database. So using '0' or '1' as > input to a boolean field is not contrary to spec, but merely outside > the spec (and in fact we will accept those strings for boolean...) The transfer into the database is regulated when you write the data into the SQL statement (as you normally would). Input through prepare/bind and input is governed by the respective client interface (e.g., embedded SQL language bindings, JDBC spec, ODBC spec). If you use libpq or psql then you're beyond all standards anyway. -- Peter Eisentraut peter_e@gmx.net
On Wed, 23 Jul 2003, Tom Lane wrote:
> Jean-Christian Imbeault <jc@mega-bucks.co.jp> writes:
> > Oops, I meant to say can someone point me to a ressource (SQL standard
> > and section?) which states that true/false must be used for booleans :)
>
> Well, there is no boolean type per se in SQL92. But there is in SQL99.
> I think the most relevant part of the spec is the definition of boolean
> literals in section 5.3:
>
> <boolean literal> ::=
> TRUE
> | FALSE
> | UNKNOWN
Was it pulled from SQL92 before it went standard? My copy of the
pre-release lists a boolean type, just like the 99 standard does.
further, intermediate SQL compliance has this in it:
24)Subclause 8.12, "<search condition>":
a) A <boolean test> shall not specify a <truth value>.
Does that mean you should only use the "is true" suntax, not the =true?
"scott.marlowe" <scott.marlowe@ihs.com> writes:
> On Wed, 23 Jul 2003, Tom Lane wrote:
>> Well, there is no boolean type per se in SQL92. But there is in SQL99.
> Was it pulled from SQL92 before it went standard? My copy of the
> pre-release lists a boolean type, just like the 99 standard does.
Where? SQL92 says nothing about a declarable boolean datatype that
I can see. They're a bit schizophrenic in that they do define a lot
of operators that are described as returning boolean ... but you cannot
create a column of type boolean, nor is there a boolean-literal construct.
AFAICS, boolean values can only exist "in flight" between operators and
a WHERE or HAVING clause in SQL92.
regards, tom lane
On Wed, 23 Jul 2003, Tom Lane wrote: > "scott.marlowe" <scott.marlowe@ihs.com> writes: > > On Wed, 23 Jul 2003, Tom Lane wrote: > >> Well, there is no boolean type per se in SQL92. But there is in SQL99. > > > Was it pulled from SQL92 before it went standard? My copy of the > > pre-release lists a boolean type, just like the 99 standard does. > > Where? SQL92 says nothing about a declarable boolean datatype that > I can see. They're a bit schizophrenic in that they do define a lot > of operators that are described as returning boolean ... but you cannot > create a column of type boolean, nor is there a boolean-literal construct. > AFAICS, boolean values can only exist "in flight" between operators and > a WHERE or HAVING clause in SQL92. Yep, you're right. I was looking at how they treated booles in search conditions, and figured they had a type to match.
<meta http-equiv="Content-Type"
content="text/html;charset=ISO-2022-JP">
I have looked around, but not found the standard. Where I have looked,
it seems to be:
t/f,
..... not........
true/flase
0/1
yes/no
Jean-Christian Imbeault wrote:
Nigel J. Andrews wrote:
I had do do this sort of thing for a some developers. It was actually 'true'
and 'false' that was wanted not the 0/1. I wrote a little plgsql function and
installed it as a cast to text.
Thanks. I know that there are workarounds, I just want to "prove" that
0/1 is not standards compliant. If I can do that I am pretty sure that
the developer will have nothing against changing to true/false.
Right not he just things postgresql must be broken because 0/1 works in
MySQL ... But he's pretty good when it comes to fixing things because
they don't follow standards. So if I can show him his SQL isn't
standards compliant he'll probably happily fix it.
Jean-Christian Imbeault
---------------------------(end of broadcast)---------------------------
TIP 3: if posting/reading through Usenet, please send an appropriate
subscribe-nomail command to majordomo@postgresql.org so that your
message can get through to the mailing list cleanly
and what about boolean attirbutes in where clauses? Is any difference between
-SELECT ... WHERE boolean_field
and
-SELECT ... WHERE boolean_field=true
?
On Wed, 2003-07-23 at 11:48, Tom Lane wrote:
-SELECT ... WHERE boolean_field
and
-SELECT ... WHERE boolean_field=true
?
On Wed, 2003-07-23 at 11:48, Tom Lane wrote:
Franco Bruno Borghesi <franco@akyasociados.com.ar> writes: > This makes me wonder, what about 't' and 'f'?... will they disappear in > newer versions of postgreSQL? No. See my comment about I/O representations being outside the spec. There is no inconsistency in the following examples: regression=# select true;bool ------t (1 row) regression=# select t; ERROR: attribute "t" not found regression=# select 't'::boolean;bool ------t (1 row) regression=# select '1'::boolean;bool ------t (1 row) regards, tom lane
Вложения
Franco Bruno Borghesi <franco@akyasociados.com.ar> writes: > This makes me wonder, what about 't' and 'f'?... will they disappear in > newer versions of postgreSQL? Nothing to do with the spec, but can I toss one comment in here: Having booleans printed as 't' and 'f' is the single most annoying thing in all of postgres. I swear it has caused more bugs in my code than any single factor. The problem is that any language driver that just treats all columns as text -- which is presumably most if not all -- will receive the column as 't' and 'f'. Both of which evaluate to true in virtually every language. If it printed booleans as 0 and 1 then pretty much every language would be able to interpret them properly. Instead I have to have =='f' and =='t' strewn throughout my code everywhere making it harder to read and extremely fragile. If I forget one anywhere I silently get subtly broken semantics. This problem might be reduced with the binary protocol since the language drivers will be able to ship boolean values in binary format and allocate variables of the appropriate type in the language. But still a lot of drivers will just bind everything as text anyways, and the goal when casting values to text should be to pick values that get interpreted appropriately. -- greg
Greg Stark wrote: > Instead I have to have =='f' and =='t' strewn throughout my code everywhere > making it harder to read and extremely fragile. If I forget one anywhere I > silently get subtly broken semantics. Why did you do that? Why not create a single function (isTrue()?) that you pass the field through and let the function evaluate it. It doesn't solve the issue, but it would make your code more stable, and if things ever changed (ie, from a "t" to a "1" or something) - you just update that one function. In fact (and my C/C++ memory is hazy) - you should be able to create a set of functions or methods in you code/classes that could take a variety of data types as the passed type, and have each of those funnel down to one type to evaluate and pass up the chain - so you could call the function isTrue(int) or isTrue(char *) or something, and it would always work OK (ie, a polymorphic function/method). That way, your code could always call the *same* function/method, and always know it was going to return the same value - but behind the scenes, it would be doing all the "funky exercises" needed to return the proper value. If it ever radically changes (or you change the DB), you change one function, and the system keeps working (no having to hunt and change 'n' lines of code). Yes, there would be a little more overhead - but future maintenance coders (or yourself, down the line) will thank you for your "foresight". Andrew L. Ayers Phoenix, Arizona -- CONFIDENTIALITY NOTICE -- This message is intended for the sole use of the individual and entity to whom it is addressed, and may contain informationthat is privileged, confidential and exempt from disclosure under applicable law. If you are not the intendedaddressee, nor authorized to receive for the intended addressee, you are hereby notified that you may not use, copy,disclose or distribute to anyone the message or any information contained in the message. If you have received thismessage in error, please immediately advise the sender by reply email, and delete the message. Thank you.