Re: fscanf

[John W. Eaton]

On 14-Apr-1999, Joao Cardoso <address@hidden> wrote:

| Thomas Walter wrote:
| > 
| > >>>>> "Joao" == Joao Cardoso <address@hidden> writes:
| > 
| >     Joao> Thomas Walter wrote:
| >     >>
| >     >> Hello,
| >     >> is it only me who has problems with 'fscanf' and friends?
| >     >>
| >     >> I have measured data in a data file written as ASCII, i.e.:
| >     >> 10  123.4
| >     >> 20  234.5
| >     >> 30  345.6
| >     >> 40  456.7
| >     >> 50  567.8
| >     >> 60  678.9
| >     >>
| >     >> This is data in one line belongs together.  In this case I have 2
| >     >> columns and the number of lines/rows is unknown, bu defined by EOF.
| >     >> If I read the data with this knowledge I always jump into the same
| >     >> pit.  The 'fscanf' function wants lines and colomuns in the opposited
| >     >> order.
| >     >>
| >     >> Apart from that the matrix I get has the size 2x6, this means:
| >     >> 10 20 30 40 50 60
| >     >> 123.4 234.5 345.6 456.7 567.8 678.9
| >     >>
| >     >> Is this behaviour only strange for me?
| 
| It seems that the doc is not updated.

Yes, the bug is in the docs.  It should read:


  [A, COUNT] = fscanf (FILENUM, FORMAT [, SIZE])

  Read from FILENUM according to FORMAT, returning the result in the
  matrix A.  SIZE is optional.  If present, it can be one of

         Inf : read as much as possible, returning a column vector
               (unless doing all character conversions, in which case a
               string is returned)
          NR : read up to NR elements, returning a column vector
    [NR, NC] : read up to NR x NC elements, returning a matrix with NR rows
   [NR, Inf] : read as much as possible, returning a matrix with NR rows

  If it is omitted, a value of Inf is assumed.

  The number of items successfully read is returned in COUNT.

  [A, B, ...] = fscanf (FILENUM, FORMAT, "C")

  Read from FILENUM according to FORMAT, with each conversion specifier
  in FORMAT corresponding to a single scalar return value.  This form is
  more `C-like', and also compatible with previous versions of Octave

Thanks,

jwe