Re: [Groff] Re: .substring bug - indicies don't work as documented(?)

[Bernd Warken]

On Sun, Oct 28, 2001 at 08:56:38AM +0000, Ralph Corderoy wrote:
> 
> ... Python ..
>
Python starts the position counting with 0, like C, while groff starts
at 1.  So Python positional calculation is fundamentally different and
cannot be taken over.

> Using zero for the end of the string and negative back from there makes
> it easy to refer to the last three characters without doing the maths
> based on string length.
> 
> > Moreover, 0 means everything up to behind the last character; so the
> > last character should be -1 anyway.
> 
> Don't quite get your point on this bit.
> 
0 is meant to be the end of the string, i.e. the position _behind_ the last
character, while the number of the last character denotes the position
on the last character.  This is comparable to the difference between
a dynamically defined '\0' terminated string and a character array of fixed 
length.

Take for example ".ds s 12345".

Then ".substring s 4 5" could be viewed as printing all characters 
from the 4th up to and including the 5th one including, so the pointer 
is at position _on_ 5 when the program terminates.

On the other hand, ".substring s 4 0" could be viewed as printing every-
thing up to the end of the string, but this end lies _behind_ the last
character.

Both command return the same result "45", but they are functionally
different. 

With the same argument, -1 means the position _on_ the last character,
because 0 means the position _behind_ it.

So ".substring s -1 -1"  means `everything from the last one to and 
including the last character', while ".substring s -1 0" means `everything 
from the last one to the end of the string'.  Tho both calls result in the 
same substring, they are functionally different.

It is much easier to use ".substring s 1 -2" in the meaning `everything
from the first up to and including the second but last character'.

> > Also 2 args in reversed order could be made into reversing the
> > output.
> 
> Yes, it could be.  I guess it depends on whether reversing the string
> or having the indices in the wrong order is more common.  This way
> avoids the `if (i1 > i2) swap(i1, i2)'.  Another option would be
> Python's way of returning an empty string which is again sometimes what
> you want.
> 
I agree to all possibilities, except the swapping.