[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Format descriptors in NUMBER->STRING ?



   Date: Fri, 29 Dec 89 03:28:42 EST
   From: "Michael R. Blair" <ziggy@hx.lcs.mit.edu>

   In the Revised 3.99 Report, why were format descriptors replaced by mere
   <radix>? That is,

    (NUMBER->STRING number radix)  vs  (NUMBER->STRING number format)

   I don't object to the left form so long as the right form is also an essential
   procedure. Why do I care?  Maybe I don't... so long as you explain to me how I
   can get the precise control over FIX/FLO/SCI that I had with formats, without
   having to write some non-trivial string frobbing procedures of my own.

   Was this ever voted on, for instance at an authors meeting which I missed,
   or was this the result of editorial oversight?

This change was requested by the IEEE editors.  The reason for the
request was that (1) it was felt that this was a baroque method for
accomplishing this, somewhat like `format' of Common Lisp, and (2) to
our knowledge, no one had implemented any of the formats except for
`heur'.

Because no one had implemented these procedures, the "precise control"
that you had with the format arguments is exactly what you have now.

It was proposed at that time that a set of formatting procedures be
implemented and submitted to the yellow pages, or perhaps to be added
to the RnRS.  There would be one procedure for each type of format,
avoiding the issue of designing a "format language" to represent the
procedure calls.