• Ben Hinkle (18/18) Mar 09 2004 Here's another proposal for an extensible stream.printf:
• J Anderson (7/26) Mar 09 2004 Good idea!
• Ben Hinkle (8/11) Mar 09 2004 Strings and arrays wouldn't be able to use the %D specifier since they
• Ben Hinkle (5/8) Mar 09 2004 oops - I meant print(..) not printf(...). And maybe the brackets

Ben Hinkle <bhinkle4 juno.com> writes:

Here's another proposal for an extensible stream.printf:

add the conversion specifier %D (for D object). This
specifier will call the object's print method. If
object.d also got a print method that accepted 
options then %D could support formats like %20D etc.

Right now stream.printf just calls the underlying 
printf with the entire format string. To support
%D stream.printf would have to go through the format
string and call the underlying printf for standard
conversions and dispatch any time it sees %D. Similar
code would work for printf-like functions like sprintf.

If this works out I'd like to change print in object.d
to just something like printf("[Object %p]",this); 
instead of the current printf("Object %p\n",this);

I don't recall seeing this proposed before but I
could be wrong. 

thoughts?

-Ben

J Anderson <REMOVEanderson badmama.com.au> writes:

Ben Hinkle wrote:

Here's another proposal for an extensible stream.printf:

add the conversion specifier %D (for D object). This
specifier will call the object's print method. If
object.d also got a print method that accepted 
options then %D could support formats like %20D etc.

Right now stream.printf just calls the underlying 
printf with the entire format string. To support
%D stream.printf would have to go through the format
string and call the underlying printf for standard
conversions and dispatch any time it sees %D. Similar
code would work for printf-like functions like sprintf.

If this works out I'd like to change print in object.d
to just something like printf("[Object %p]",this); 
instead of the current printf("Object %p\n",this);

I don't recall seeing this proposed before but I
could be wrong. 

thoughts?

-Ben
  

Good idea!

And if a D string is being used it would be able to format that properly 
without using %.*s.  It could also do an array of objects (since in D 
there is a distinction between arrays and pointers).

-- 
-Anderson: http://badmama.com.au/~anderson/

"Ben Hinkle" <bhinkle4 juno.com> writes:

 And if a D string is being used it would be able to format that properly
 without using %.*s.  It could also do an array of objects (since in D
 there is a distinction between arrays and pointers).

Strings and arrays wouldn't be able to use the %D specifier since they
aren't objects and don't have a "print" method. Using %s for D strings
is possible but then it wouldn't work for C strings. So in that case
to print a C string you call std.c.printf("%s",...) and to print a D string
you call stdout.printf("%s",...). That doesn't look too bad to me, though
I haven't thought about the details. Printing arrays of arbitrary types is
probably best left to the user.

-Ben

"Ben Hinkle" <bhinkle4 juno.com> writes:

 If this works out I'd like to change print in object.d
 to just something like printf("[Object %p]",this);
 instead of the current printf("Object %p\n",this);

oops - I meant print(..) not printf(...). And maybe the brackets
[Object %p] isn't that great since it might look like an array.
I haven't thought too much about object.print vs object.printf...

Walter, why did you choose object.print? I assume it's because
printf takes a format string (hence the "f" in printf).