Re: Question about bin2dec

[John W. Eaton]

On 29-Jun-2006, Muthiah Annamalai wrote:

| > I have some questions regarding the bin2dec function.  My understanding 
| > is that it converts a string of binary characters into a decimal 
| > number.  However, it looks to me that spaces within the string give 
| > rise 
| > to a completely different result.  As an example:
| >
|  >> x = [1 1 1 1 0];
|  >> x_str = int2str(x);    %convert to string
| > fmt = %1d%3d%3d%3d%3d
| 
|  >> x_str
| > x_str = 1  1  1  1  0
|  >> bin2dec(x_str)
| > ans = 4680
| 
| > However, if I type
|  >> bin2dec('11110')
| > I get an answer of 30, which is what I would expect.  The spaces 
| > inbetween the characters throw the answer off from the expected result.
| 
| > Is this a bug, or a "feature"?  It seems to me that we would not want 
| > the result from the first method.
| > -Ron
|    
|   I think that is a 'feature'. Especially you can use it like this:
|   x=sprintf("%d",x);
|   bin2dec(x)
|    
|   which will do the job.

It is also a compatibility bug, though the behavior appears to have
been a deliberate choice.  Here is the code from base2dec.m:

  ## Right justify the values before anything else.
  d = strjust (d, "right");

  ## Lookup value of symbols in symbol table, with invalid symbols
  ## evaluating to NaN and space evaluating to 0.
  table = NaN * ones (256, 1);
  table (toascii (symbols (1 : base))) = 0 : base-1;
  table (toascii (" ")) = 0;
  d = reshape (table (toascii (d)), size (d));

Should we opt for compatility here?  I don't see that converting
spaces to 0 is any better than ignoring them.

jwe