Re: fprintf compatibility

[John W. Eaton]

On 31-Dec-2002, Paul Kienzle <address@hidden> wrote:

| John W. Eaton wrote:
| 
| >What do you think about this feature of Matlab?
| >
| >  >> fprintf ('%s\n', [111; 100; 100]);
| >  odd
| >
| >I know that Matlab (originally, anyway, though perhaps not anymore)
| >stores strings as double precision matrices with a flag set saying to
| >interpret the numbers as ASCII and print them as strings, etc.  But
| >the matrix here is not even tagged as a string!
| >
| >So, should Octave copy this bug^H^H^Hfeature?
| >
| Won't this happen already with implicit_num_to_str_ok = 1?
| 
| Okay, no it doesn't.

Right, but perhaps it should.

| Even if it did, shouldn't it output the following:
| 
| o
| d
| d
| 
| I have nothing in particular against implicit char->num/num->char, but I 
| also don't
| mind putting setstr() around the matrices.

Setstr is definitely the quick fix.

| I'm a little disappointed that
| sprintf('%s\n',['o';'d';'d']) doesn't print a column though.

Disappointing or not, that part is already compatible.  :-/  But it
does seem inconsistent with the behavior of

  fprintf ('%d\n', [1; 2; 3])

for example.

jwe