       • To: mathgroup at smc.vnet.net
• Subject: [mg64584] Re: [mg64551] Increment and AddTo
• From: Sseziwa Mukasa <mukasa at jeol.com>
• Date: Thu, 23 Feb 2006 00:34:22 -0500 (EST)
• References: <200602221058.FAA23452@smc.vnet.net>
• Sender: owner-wri-mathgroup at wolfram.com

```On Feb 22, 2006, at 5:58 AM, dh wrote:

> Hello,
> can anybody explain the following behaviour:
> \$Version = 5.1 for Microsoft Windows (October 25, 2004)
>
> i = 0;
> v = {1, 2, 3};
> v[[++i]] += 10;
> v
> i
>
> gives
> {1,11,3}
> 2
>
> i has been incremented by 2!

Using Trace the reason should be clear.  Trace gives:

Trace[v[[++i]] += 10;]

{v[[++i]]+=
10;,{v[[++i]]+=
10,{{v,{1,2,3}},{++i,{i,0},{i=1,1},
1},{1,2,3}[],1},{{1+10,11},
v[[++i]]=11,11},11},Null}

Since x+=dx is explicitly equivalent to x=x+dx it should be clear
that x will be evaluated at least once before the assignment is done,
that results in the expression x+dx evaluating to 11 in this case.
Set[Part[_,_]] has special semantics though, the second argument of
Part has to be evaluated before the assignment to know where the
substitution must be made.  Since Set has the attribute HoldFirst the
second argument to Part is ++i not 1 so ++i gets evaluated again as
you can see in the Trace where the expression v[[++i]]=11 gets
evaluated.  At that point ++i has already been evaluated once so
obviously i will be incremented twice.

> i++ is even more interesting:
> i = 1;
> v = {1, 2, 3};
> v[[i++]] += 10;
> v
> i
>
> gives:
> {1,11,3}
> 2

It's pretty much the same story, i++ gets evaluated twice, however
the final value of i which I get is 3 not 2 which makes sense.

> Further:
> i = 0;
> v = {1, 2, 3};
> ++v[[++i]] ;
> v
> i
>
> gives:
> {1,2,3}
> 2
> Again i has been incremented by 2, but v has not been touched at all.

++ is not equivalent to a Set expression so ++v[[++i]] does not share
the special semantics that Set and Part do and the result of the
expression does not modify the value of anything.

Regards,

Ssezi

```

• Prev by Date: Re: Counting the number of paths in a network
• Next by Date: Re: General--Making the DisplayFormula style in ArticleModern look like Traditional