> mysql> select x.a from test x where 1;
>
> 1 evaluates to be true, so this should select all rows. And it does.
>
> 5 rows in set (0.00 sec)
>
> mysql> select * from test where 0;
>
> This evaluates to false, so this should not select any rows. And it does.
>
> Empty set (0.00 sec)
>
> mysql> select x.a from test x where x.a;
>
> This should select all rows where the a column's value evaluates to true.
> And it does.
>
> 3 rows in set (0.00 sec)
The tree cases above is a typical C/C++ behavior, so nothing is wired here.
But the next one does not. (Rather it behave like a typical C/C++ fault in
handling/checking integrity/consistency of pointers somewhere.)
> mysql> select x.a from test x where 'a';
> Empty set (0.00 sec)
>
> I'm not quite sure why this doesn't return anything, to tell the truth.
Neither I, it seams strange. This value has to be stored in some kind
of pointer to an object or char * in the program it self. I would first
assume that the content of the pointer is tested, but that cant be true
since the content in this case is the number 65.
So then you would suspect that they test a pointer instead. But that
does not make sence either. Thre pointer should exists, right?
But maybe there is something really strange going on with the
interpretation of the SQL query insteed?
I kind of recognize this situation with the BASIC interpreted built
in the system I did work with before. One could build really funny
(working) BASIC statement due to that the interpreter went confused
with the sysntax sometimes.
I don't know how MySQL tokenizes the SQL query, but if it breaks up
the token in a certain way, then it is possible that the interpreter
finds the string 'a', and then assumes it MUST have an operator, but
the operator might then point to NULL, an dthen nobody know what
happens after that, except that the end results evaluates to zero.
Hmmm... this is indeed interesting to know what really happens, maybe
I should download the source code and stepdebug just top see what actually
happens. But that has to be a later project...
>That
> said, I cannot understand why you'd ever set up a query like these.
I did actually try to provoke a 'syntax error' message will all test
cases, but they failed. However I learnt something important from this:
If an applications programmer does something like this:
SELECT <whatever>
FROm <a bunch of tables>
WHERE <lots of primary key lookup between tables>
<and a bunch of nested SELECT statements>
AND Attribute
/* and the app programmer took a coffee break here, and
** when returned forgot to add "= <some hardcoded value>"
*/
<lots of other junk>
Since the query compiler does not generates a syntax error here, this might
be a source of introducing faults in the system. That's why it is important
to know exactly how the system behaves. No I know this, but is there other
surprises that I would like to know about?
> On Tuesday 16 April 2002 11:21 am, Svensson, B.A.T. (HKG) wrote:
> > Basically I am just curious, and I did perform some few small test.
> > But, I was suspecting to get all rows when I did something like
> > "WHERE <valid string>", but had to write something like "WHERE NOT
> > <valid string>" which feels a little bit upside down for me considering
> > that integers apart from zero evaluates to true.
> >
> > Also it seams like this is not ISO SQL, so it could be a bit tricky
> > to predict the outcome without actually having some specification of
> > the behavior.
> >
> > //Anders
> >
> > > -----Original Message-----
> > > From: Christopher Thompson
> > > Sent: Tuesday, April 16, 2002 6:24 PM
> > > To: Svensson, B.A.T. (HKG); Mysql List
> > > Subject: Re: WHERE codition test
> > >
> > >
> > > I'm not sure what you are asking. Everything is working correctly as far
> > > as I can see.
>
---------------------------------------------------------------------
Before posting, please check:
http://www.mysql.com/manual.php (the manual)
http://lists.mysql.com/ (the list archive)
To request this thread, e-mail <[EMAIL PROTECTED]>
To unsubscribe, e-mail <[EMAIL PROTECTED]>
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
