Re: Sequential evaluation of lists
- To: mathgroup at smc.vnet.net
- Subject: [mg53091] Re: [mg53076] Sequential evaluation of lists
- From: William K Reinholtz <kirk.reinholtz at jpl.nasa.gov>
- Date: Thu, 23 Dec 2004 07:58:31 -0500 (EST)
- References: <200412220953.EAA04525@smc.vnet.net>
- Sender: owner-wri-mathgroup at wolfram.com
Ray:
I don't think you missed anything, but still I think there's a reason
for the behavior you note.
The documentation rightly tells you to not depend upon the order of
evaluation for two reasons: (1) It would be difficult or impossible for
Mathematica to do the various transformations it performs if such an
assumption were made; and (2) It would restrict the implementation of
the kernel unduly.
The actually Mathematica kernel does not enforce the ordering (e.g. by
rejecting your example below, or doing it in parallel) because that
would be quite costly to implement (if not impossible to implement) and
there are probably many practical difficulties that would follow from
such an implementation that would make Mathematica less useful and less general.
And so it seems a pragmatic decision to not check for violations and
not to execute in parallel: it makes for a more useful product overall.
On Dec 22, 2004, at 1:53 AM, Ray Koopman wrote:
> When I first started using Mathematica (v2), one of the features that
> I found rather surprising is its sequential evaluation of lists, as in
>
> In[1]:= x = 0; {x++,x++,x++}
> Out[1]= {0,1,2}
>
> I had expected a warning that such code should be avoided because
> it presumed sequential evaluation, which could not be guaranteed,
> and a recommendation to treat list elements as being evaluated in
> parallel -- if not simultaneously then in no particular order.
> However, so far I have found no exception to sequential evaluation
> and no mention of it in any documentation. Have I missed something?
>
- References:
- Sequential evaluation of lists
- From: "Ray Koopman" <koopman@sfu.ca>
- Sequential evaluation of lists