bug#16293: 24.3.50; doc of `string-to-number'

bug#16293: 24.3.50; doc of `string-to-number'

[Drew Adams]

Say what happens if STRING cannot successfully be parsed as a number:
the function raises an error.

The doc (e.g. manual) describes the parsing in a fair amount of detail,
and it describes how the result can end up a floating-point number if
the integer value would be too large.  But it says nothing about what
happens if no number can be identified for the string.



In GNU Emacs 24.3.50.1 (i686-pc-mingw32)
 of 2013-12-27 on ODIEONE
Bzr revision: 115778 rgm@gnu.org-20131228000456-1797o8z6veuyozs0
Windowing system distributor `Microsoft Corp.', version 6.1.7601
Configured using:
 `configure --prefix=/c/Devel/emacs/binary --enable-checking=yes,glyphs
 'CFLAGS=-O0 -g3' LDFLAGS=-Lc:/Devel/emacs/lib
 CPPFLAGS=-Ic:/Devel/emacs/include'

bug#16293: 24.3.50; doc of `string-to-number'

[Jay Belanger]

> Say what happens if STRING cannot successfully be parsed as a number:
> the function raises an error.

When would it give an error, rather than 0?

bug#16293: 24.3.50; doc of `string-to-number'

[Drew Adams]

> > Say what happens if STRING cannot successfully be parsed as a number:
> > the function raises an error.
> 
> When would it give an error, rather than 0?

If it never raises an error, then say explicitly that for any
string that cannot successfully be parsed as a number it returns 0.

(Cf. bug #16294, which was giving me the impression at first that
`string-to-number' was raising an error.)

bug#16293: 24.3.50; doc of `string-to-number'

[Drew Adams]

> >> > Say what happens if STRING cannot successfully be parsed as a number:
> >> > the function raises an error.
> >>
> >> When would it give an error, rather than 0?
> >
> > If it never raises an error, then say explicitly that for any
> > string that cannot successfully be parsed as a number it returns 0.
> 
> It does say:
>      If the first character after the ignored whitespace is neither a
>      digit in the given base, nor a plus or minus sign, nor the leading
>      dot of a floating point number, this function returns 0.
>
> Should this be reworded, or perhaps moved to earlier in the
> documentation?

It should be corrected to be more complete.  It's not just about the
first character.  Any string of characters that cannot be successfully
parsed as a number results in a return value of zero (apparently - I
have not checked the code).  "+@", for instance.
 

bug#16293: 24.3.50; doc of `string-to-number'

[Paul Eggert]

Thanks for the bug report; fixed in trunk bzr 115868.