Zero - Revision history

Stefan Haubenthal: /* Comparing variables against zero (0) */

2021-02-19T15:11:11Z

Comparing variables against zero (0)

← Older revision		Revision as of 15:11, 19 February 2021
Line 30:		Line 30:
	This is not a problem in Extended BASIC (any version).		This is not a problem in Extended BASIC (any version).

	The difficulty is caused by the multiple ways that the radix 100 floating point system can represent a zero value. In particular, internally the comparison to zero (and only zero) should be made with ''only'' the first two bytes of a number. By changing the comparison to be with something other than 0, all bytes are compared and we do not have a problem.		The difficulty is caused by the multiple ways that the [[radix-100]] floating point system can represent a zero value. In particular, internally the comparison to zero (and only zero) should be made with ''only'' the first two bytes of a number. By changing the comparison to be with something other than 0, all bytes are compared and we do not have a problem.

	Call it a bug or a feature, but if you program in TI BASIC you need to be aware of this. It also has the potential to cause a problem if you use radix 100 floating point in any other language.		Call it a bug or a feature, but if you program in TI BASIC you need to be aware of this. It also has the potential to cause a problem if you use radix 100 floating point in any other language.

Stefan Haubenthal: /* Comparing variables against zero (0) */ typo

2018-06-10T06:46:37Z

Comparing variables against zero (0): typo

← Older revision		Revision as of 06:46, 10 June 2018
Line 23:		Line 23:
	Now amend line 110 to be:		Now amend line 110 to be:
	<pre>		<pre>
	~~100~~ IF A+1<>1 THEN 100		110 IF A+1<>1 THEN 100
	</pre>		</pre>

Stephen Shaw: Add note re TI Basic bug when 0 isn't really 0

2017-12-24T21:01:44Z

Add note re TI Basic bug when 0 isn't really 0

New page

==Comparing variables against zero (0)==

There is a bug which causes a problem if you program in TI BASIC:

If you use the split keyboard as follows:
<pre>
100 CALL KEY(1,A,B)
110 IF A<>0 THEN 100
120 PRINT "YOU PRESSED KEY X"
</pre>

If you press key X, using the split keyboard as above (the same is true with keyboard 2 and key M), you should expect the value of A to be zero and the program to go to line 120.

It doesn't.

To check the value of A really does go to zero try inserting another line:
<pre>
105 PRINT A
</pre>

When you press X now you see the value zero appear on screen but the comparison still fails.

Now amend line 110 to be:
<pre>
100 IF A+1<>1 THEN 100
</pre>

Mathematically the two lines 110 are equivalent, but in TI BASIC they do not behave the same.

This is not a problem in Extended BASIC (any version).

The difficulty is caused by the multiple ways that the radix 100 floating point system can represent a zero value. In particular, internally the comparison to zero (and only zero) should be made with ''only'' the first two bytes of a number. By changing the comparison to be with something other than 0, all bytes are compared and we do not have a problem.

Call it a bug or a feature, but if you program in TI BASIC you need to be aware of this. It also has the potential to cause a problem if you use radix 100 floating point in any other language.