Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!agate!shelby!csli!poser
From: poser@csli.Stanford.EDU (Bill Poser)
Newsgroups: comp.lang.c
Subject: Re: integer to string conversion
Keywords: code
Message-ID: <9814@csli.Stanford.EDU>
Date: 24 Jul 89 10:14:09 GMT
References: <3020@ohstpy.mps.ohio-state.edu> <9813@csli.Stanford.EDU> <18702@mimsy.UUCP>
Sender: poser@csli.Stanford.EDU (Bill Poser)
Reply-To: poser@csli.stanford.edu (Bill Poser)
Organization: Center for the Study of Language and Information, Stanford U.
Lines: 18

It is correct that the routine I posted doesn't do the right thing on the
most negative integer. That requires a special case, which I left out.

Chris Torek is correct that a binary search will speed it up even more,
but as he says that means knowing the range of integers on the machine,
which reduces portability. The simplest and most portable way (across
environments where sprintf is available) to do this
is to use sprintf, the fastest and least portable is to write an
assembly language routine that takes advantage of the machine's
instruction set and knowledge of the size of integers. The routine
I provided is the middle ground - it gives a speed advantage over
sprintf (at least on certain common processors) but remains pretty portable.

I think Chris is being cranky about the need for the routine.
I began with the most straightforward solution (sprintf), gave a routine
that shows how the conversion is done, and was explicit about the
magnitude of its speed advantage and how it might vary. What more can you
ask for?