Fabio,
It's a bug as agreed. Why would I ever reindex it. Assuming there might be a need, yes reindex kills the index but still you can give a kiss and bring it back to life in 2-3 lines of code. It seems more cumbersome to me to be carefull to add extra space.
Cetin

>>Thomas,
>>
>>That's avoiding the index. And that's what I was doing, ie using (j_ref=0) instead of (bintoc(j_ref) == bintoc(0))
>>
>>I'm currently testing Cetin's suggestion and I think, although not confirmed yet, that it solves the problem
>>
>>cheers,
>
>Gregory,
>move the index don't fix the bug.
>The first REINDEX will kill the index.
>
>Run the code with this workaround:
>

>	clear
>	set ansi on
>	set exact on
>	set collate to 'machine'
>	set optimize on
>
>	create cursor CursorTest ;
>		( 	j_ref		I ;
>		)
>	
>	index on bintoc(j_ref)+' ' tag j_ref
>	
>	insert into CursorTest values(44 )
>	insert into CursorTest values( 0 )
>	insert into CursorTest values(44 )
>
>	REINDEX
>	? KEYMATCH(bintoc(0)+' ')
>
>	?'-------------------------------------'
>
>	locate for (bintoc(j_ref)+' ' == bintoc(0)+' ')
>	?'optimize locate 0 (TRUE)', found(), '  ERROR'
>
>	locate for (bintoc(j_ref)+' ' == bintoc(0)+' ') NOOPTIMIZE
>	?'no optimize locate 0 (TRUE)', found(), ' OK'
>
>	local aa[1]
>	aa[1] = 0
>
>	select count(*) ;
>		from CursorTest ;
>		into array aa ;
>		where (bintoc(j_ref)+' ' == bintoc(0))
>
>	?'optimize count 0 (1):', aa[1], ' ERROR'
>
>	set optimize off
>	select count(*) ;
>		from CursorTest ;
>		into array aa ;
>		where (bintoc(j_ref)+' ' == bintoc(0))
>	set optimize on
>
>	?'no optimize count 0 (1):', aa[1], aa[1], ' ERROR'
>
>
>	locate for (bintoc(j_ref)+' ' == bintoc(44)+' ')
>	?'optimize locate 44 (TRUE), recno 1:', found(), 'OK ', recno()
>
>	insert into CursorTest  values( 0 )
>	locate for (bintoc(j_ref)+' ' == bintoc(0)+' ')
>	?'optimize locate 0 (TRUE), recno 2:', found(), recno(), 'ERROR'
>
>

>
>Fabio