Precision in calculations MySQL - mysql

Question

I have this strange problem. I have a table containing 2 numbers, but a simple calculation makes a precision fault:
mysql> create table tblX as select '2.7' as H, '2100' as H2;
Query OK, 1 row affected (0.32 sec)
Records: 1 Duplicates: 0 Warnings: 0
mysql> select *, h-h2/1000 from tblX;
+-----+------+--------------------+
| H | H2 | h-h2/1000 |
+-----+------+--------------------+
| 2.7 | 2100 | 0.6000000000000001 |
+-----+------+--------------------+
1 row in set (0.00 sec)
This example recreates a problem I have in a setup, where formulas, field and their values are extracted in a way where some fields have to be varchars but others are used in calculations as above. So the formula can not be changed. Another example could be a formula like: if(code='c' and h-h2/1000>0.6,h,0)
where the fields are created by a pivot operation.
My assumption is, that implicit conversion of varchar to decimal causes this problem. My question is: Are there any server settings that could change this implicit conversion. Or are there any other solution for this.
EDIT:
If I modify the columns like this:
alter table tblX MODIFY column h decimal(20,4), MODIFY h2 decimal(20,4);
The result will be correct. But I do not know the names of the columns in advance as they are created on the fly in temporary tables. It would be nice if it was possible to create a script that modify all columns that is only containing numeric values.

Answer 1

I am not aware about any server settings. But you can do explicit conversion in the sql as well.
Try something like this :
select *, CAST(h-h2/1000 AS DECIMAL(10,2)) from tblX;
Refer MySQL documentation. https://dev.mysql.com/doc/refman/5.5/en/fixed-point-types.html

Answer 2

The problem is, that varchars in a formula apparently implicit converts to floats and not decimals. Changing the formulas is not an option in this case. But modifying column datatype is an option.
The solution is to make a stored procedure that loops thru the relevant columns and modify them to a decimal datatype.
The table that contains the incorrect datatypes is named tmp.
I made a temporary table called tmpPar containing all column names that take part in a formula as numeric values. Then I set up a cursor and loop thru all those columns. Cursor loop looks like this:
DECLARE a varchar(255);
DECLARE cur1 CURSOR FOR SELECT CName FROM tmpPar;
...
Open cur1;
simple_loop: LOOP
fetch cur1 into a;
IF no_more_rows=1 THEN
LEAVE simple_loop;
END IF;
set #sql = null;
set #sql = concat('alter table tmp MODIFY column ' , a , ' decimal(20,4)');
PREPARE stmt FROM #sql;
EXECUTE stmt;
END LOOP simple_loop;
Close cur1;