MathGroup Archive 2011

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: Span Set nasty bug

  • To: mathgroup at smc.vnet.net
  • Subject: [mg123880] Re: Span Set nasty bug
  • From: Oliver Ruebenkoenig <ruebenko at wolfram.com>
  • Date: Mon, 26 Dec 2011 17:30:46 -0500 (EST)
  • Delivered-to: l-mathgroup@mail-archive0.wolfram.com
  • References: <201112261153.GAA14434@smc.vnet.net>


On Mon, 26 Dec 2011, Shizu wrote:

> Hello there,
>    I think Roby's logic is right.  Oliver's suggestion makes both results wrong -- t[[1,1]] should not be set to 1.
>    The rule ("TableCompileLength" -> 1) is not documented.  Span "1;;0" is not documented either.

This was to illustrate the the behavior can be changed. This is an issue 
with packed arrays. So unpacking (or setting "TableCompileLength" -> 
Infinity) using Developer`FromPackedArray is a better option.

>    It's not safe to use something that is not documented.
>

That is right, you should be cautious.

As Rob mentioned this is with tech support now and is investigated.

Oliver

>    You may want to invent your own version of Set[] for matrices.
>
>



  • Prev by Date: AbsoluteOptions and NDSolve
  • Next by Date: Re: arrows in the parametric plot
  • Previous by thread: Re: Span Set nasty bug
  • Next by thread: Approximating ratio of incomplete gamma functions