Thanks for reviewing Eli.

> If we can implement DFP printing in GDB, why should we also check for
> native DFP support and use that if available?  Are there any
> advantages in the native DFP printing, and if so, what are they?

Yes. With native DFP support we would have access to all the modifiers
(precision, width etc) available for printf, whereas, in the string
conversion solution we would just have a straight value-to-string
conversion mechanism without any additional capability.

> Also, if this patch is accepted, please submit also a suitable patch
> for the manual (doc/gdb.texinfo), where we describe the format
> conversions supported by the `printf' command (node "Output").

No problem. I'll supply that patch.

> A few questions/comments to the patch itself:
> 
> > -	double_arg, long_double_arg
> > +	double_arg, long_double_arg, decfloat_arg
> 
> You are using decfloat_arg unconditionally, but I don't see it defined
> anywhere in today's CVS.  Am I missing something?

decfloat_arg is being defined in an "enum" structure together with the
other values (double_arg, long_double_arg, int_arg etc). Is there a
different place you think it needs to be defined as well?

> +	      /* Replace %H, %D and %DD for %s's.  */
> Did you mean "Replace ... _with_ %s's"?
Fixed.

Regards,
Luis