MathGroup Archive 2000

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

Search the Archive

buggy part allocation ?

  • To: mathgroup at smc.vnet.net
  • Subject: [mg23443] buggy part allocation ?
  • From: slinger at signal.dera.gov.uk (Chris Slinger)
  • Date: Wed, 10 May 2000 02:32:23 -0400 (EDT)
  • Sender: owner-wri-mathgroup at wolfram.com

I'm having problems with silly results.  The cause seems to be faulty
parts allocation.  This is illustrated below.  I've mailed wolfram
support but have yet to receive a reply.  I'm running v4.0.1 for X on
Sun Solaris.  I know that there has been trouble with Range[] in the
past, but thought that this had been fixed.

Can anyone explain the following behaviour ?

In[1]:=$Version
Out[1]="4.0 for Solaris (July 21, 1999)"

In[2]:=$CreationDate
Out[2]={1999, 7, 21, 1, 43, 0}

In[3]:= m = 256; list = Table[0, {m}, {m}];
insert = Table[Random[Complex, {-1 - I, +1 + I}], {10}, {250}];

In[4]:= list[[Range[10], Range[250]]] = 1*insert; MatrixQ[list]
Out[4]=True

In[5]:=list[[Range[10], Range[250]]] = 2*insert; MatrixQ[list]
Out[5]=False

In[6]:= thing = 2*insert; list[[Range[10], Range[250]]] = thing; MatrixQ[list]
Out[6]=False

__________________________

Out[5] and Out[6] should, I imagine, be "True" - simply changing from
1*insert to 2*insert breaks the parts allocation.

The above behaviour is also exhibited on NT systems.  What is happening
here ?  Am I doing something daft ?!

Please copy any correspondence on this to my e-mail, as I often
miss the newsgroup postings....

Thanks,
-chris
slinger at signal.dera.gov.uk

PS - this has echoes of the complex packed array bug (now supposed to
be fixed).  If "insert" is Real, there are no problems in the above...


  • Prev by Date: How to Play Sound from FrontEndTokenExecute
  • Next by Date: Re: Symbols & Legend Fonts in MultipleListPlot...
  • Previous by thread: How to Play Sound from FrontEndTokenExecute
  • Next by thread: Re: buggy part allocation ?