Re: integer to string conversion

Thus spake Richard Wossal ([email protected]):

> I was going to ask you about your opinion whether or not
> the non-standard itoa-family of functions have a place in
> dietlibc, when I discovered that the dietlibc features
> __ltostr (and derivates). I've never seen that before,
> Internet says they come from HP-UX.

Uh, we invented that interface ourselves. It's not for public
consumption.  The official way to do it is to use sprintf, which sucks,
obviously.

So my recommendation is to use fmt_* from libowfat
(www.fefe.de/libowfat/)

> Now what I would like is a function family in the
> dietlibc that looks like the following:
>         int itoa(int n, char *buf, int base)
> returning the length of n in characters.

> Do you think, a libc is the right place for those?
> After all, they are not "less standard" than the
> __ltostr functions, are they?
> And (on a side note), why doesn't embutils use
> __ltostr?

__ltostr is not portable at all, it's only an internal function used by
sprintf in dietlibc.  Do not use it, even if you checked for it's
existance!

As a rule of thumb: if the function name starts with __, stay away from
it :-)

A function like that does not belong in libc, because there is a
standard that says what libc should export.  Actually, there is more
than one.  But it's standardized.  What good is standardization if you
ignore it and add stuff?

Felix