negative integer division

[Imbaud Pierre]

integer division and modulo gives different results in c and python,
when negative numbers
are involved. take gdb as a widely available c interpreter
print -2 /3
0 for c, -1 for python.
more amazing, modulos of negative number give negative values! (in c).
from an algebraic point of view, python seems right, but I thought
python conformity to the underlying c compiler was a strong commitment,
obviously not here, and I found no explanation whatsoever in python doc.
no actual programming challenge for me here, I just had this bug, after
confidently translating from python to c.

[Skip Montanaro]

Re: negative integer division


Imbaud> integer division and modulo gives different results in c and
Imbaud> python, when negative numbers are involved.

http://www.python.org/doc/faq/programming.html#why-does-22-10-return-3

Skip

[Robert Kern]

Re: negative integer division

Imbaud Pierre wrote:[color=blue]
> integer division and modulo gives different results in c and python,
> when negative numbers
> are involved. take gdb as a widely available c interpreter
> print -2 /3
> 0 for c, -1 for python.
> more amazing, modulos of negative number give negative values! (in c).
> from an algebraic point of view, python seems right, but I thought
> python conformity to the underlying c compiler was a strong commitment,
> obviously not here, and I found no explanation whatsoever in python doc.
> no actual programming challenge for me here, I just had this bug, after
> confidently translating from python to c.[/color]

I don't think there's much *commitment* to follow the behaviour of your
C compiler. It's more of a default stance when the issues get too
complicated for Python to implement itself. Thus, much of the floating
point behaviour defaults to the behaviour of your platform because
floating point math is hard and getting consistent behaviour on lots of
platforms is really hard.

Integers are a piece of cake, relatively speaking, and I think that it's
worth fixing up a few warts from the C behaviour.

As for documentation, see

http://www.python.org/doc/current/lib/typesnumeric.html

--
Robert Kern
rkern@ucsd.edu

"In the fields of hell where the grass grows high
Are the graves of dreams allowed to die."
-- Richard Harter

[Scott David Daniels]

Re: negative integer division

Imbaud Pierre wrote:[color=blue]
> integer division and modulo gives different results in c and python,
> when negative numbers
> are involved. take gdb as a widely available c interpreter
> print -2 /3
> 0 for c, -1 for python.
> more amazing, modulos of negative number give negative values! (in c).
> from an algebraic point of view, python seems right, but I thought
> python conformity to the underlying c compiler was a strong commitment,
> obviously not here, and I found no explanation whatsoever in python doc.
> no actual programming challenge for me here, I just had this bug, after
> confidently translating from python to c.
>[/color]
If you read the C standard, you can see that the C compiler is free to
make -1 / 3 either -1 or 0 (but it must describe what it does). The
decision is usually to do whatever the underlying CPU does. The
rationale is that often you will be dividing positive by positive,
and you don't want the compiler sprinkling around a lot of test code.

While the difference in compiled C is a substantial time penalty
(DIVIDE vs. DIVIDE; COMPARE; JUMPGE), in Python the trade-off is
adding two instructions to many (50? 100?). Python doesn't really
mind paying for a few extra instructions to get the canonical value,
and the guarantee simplifies the user's code.

--Scott David Daniels
Scott.Daniels@A cm.Org

[Mark Jackson]

Re: negative integer division

Imbaud Pierre <pierre.imbaud@ wanadoo.fr> writes:[color=blue]
> integer division and modulo gives different results in c and python,
> when negative numbers
> are involved. take gdb as a widely available c interpreter
> print -2 /3
> 0 for c, -1 for python.
> more amazing, modulos of negative number give negative values! (in c).
> from an algebraic point of view, python seems right, but I thought
> python conformity to the underlying c compiler was a strong commitment,[/color]

AIUI the C standard is silent on the issue, and hence the C behavior is
implementation-dependent. Anyway back in 2000 I found and fixed a
Y2K-related problem in an open-source C program (xvtdl) which was down
to precisely this misbehavior. While diagnosing the problem I
implemented the algorithm in Python for test purposes, and was led
astray for a while by the fact that it *didn't* fail!

A: 42

Q: What multiple of 7 did I add to the critical expression in the Zeller
algorithm so it would remain nonnegative for the next few centuries?

--
Mark Jackson - http://www.alumni.caltech.edu/~mjackson
People who write obscurely are either unskilled in writing
or up to mischief. - Sir Peter Medawar

[Jive Dadson]

Re: negative integer division

Python does it right. C is allowed to do it anyway it likes, which was
a stupifyingly horrible decision, IMHO.

Way back when, there was a language named Pascal. I lobbied the Pascal
standards committee to define the modulus operator correctly, which they
eventually did. To my astonishment, they proceded to define the
division operator backwards to the modulus operator, so that division
and mod did not play together correctly. Duh.

[Dan Bishop]

Re: negative integer division

Mark Jackson wrote:[color=blue]
> Imbaud Pierre <pierre.imbaud@ wanadoo.fr> writes:[color=green]
> > integer division and modulo gives different results in c and[/color][/color]
python,[color=blue][color=green]
> > when negative numbers
> > are involved. take gdb as a widely available c interpreter
> > print -2 /3
> > 0 for c, -1 for python.
> > more amazing, modulos of negative number give negative values! (in[/color][/color]
c).[color=blue][color=green]
> > from an algebraic point of view, python seems right, but I thought
> > python conformity to the underlying c compiler was a strong[/color][/color]
commitment,[color=blue]
>
> AIUI the C standard is silent on the issue, and hence the C behavior[/color]
is[color=blue]
> implementation-dependent.[/color]

The *original* C standard is silent on the issue, but the current
standard requires the wrong behavior.

[Mike Meyer]

Re: negative integer division

Jive Dadson <jdadson@yahoo. com> writes:
[color=blue]
> Python does it right. C is allowed to do it anyway it likes, which was
> a stupifyingly horrible decision, IMHO.[/color]

C only does it wrong if you think that C is a high level language. It
isn't - it's a portable assembler. As such, low level things (like
this, or what happens on integer overflow, or ...) are left up to the
implementation, so it can do what's most natural for the underlying
hardware. This means that when you don't care - which I'd argue is
most of the time - you get the fastest thing the machine will do. When
you do care, you have to take care of it yourself. Of course, if you
care, you probably shouldn't be writing in assembler, you should
probably be writing in a high level language - which will make sure
the low level things get done right, irregardless of what the
underlying machine does.

Now, I'll agree with you if you want to argue that some machines do
negative integer division in stupifyingly horrible ways.

<mike
--
Mike Meyer <mwm@mired.or g> http://www.mired.org/home/mwm/
Independent WWW/Perforce/FreeBSD/Unix consultant, email for more information.

[Jive Dadson]

Re: negative integer division



Mike Meyer wrote:[color=blue]
>
> Jive Dadson <jdadson@yahoo. com> writes:
>[color=green]
> > Python does it right. C is allowed to do it anyway it likes, which was
> > a stupifyingly horrible decision, IMHO.[/color]
>
> C only does it wrong if you think that C is a high level language.[/color]

I didn't say it does it wrong. I said it does it anyway it likes --
maybe right, maybe wrong. There *is* a right way, IMHO. Python does it
that way.
[color=blue]
> [C] isn't - it's a portable assembler.[/color]

I've heard that many times, but it makes no sense to me. By definition,
the syntax of an assembly language closely resembles the format of
individual hardware instructions for a particular processor. An
assembler assembles individual hardware instructions. Back in the day,
Unix (written largely in C) and Steve Johnson's pcc (the *portable* C
compiler) together constituted a big leap forward. Implementing Unix on
new processors was infinitely easier than porting OS's written in
assembly language. So I disagree on two counts: C code is not entirely
portable, (for example, division and mod may not work the same on two
different machines), and a C compiler is most certainly not an
assembler.

[color=blue]
> Now, I'll agree with you if you want to argue that some machines do
> negative integer division in stupifyingly horrible ways.[/color]

That's why I think it was a stupifyingly horrible decision.
Understandable, but in the end an s.h.d. nonetheless. It would have
been okay to define / and % correctly, in the mathematical sense, but
also provide functions quick_div() and quick_mod() that were guaranteed
to work correctly only when both arguments were positive. The way they
did it is just too error prone, akin to early optimization. It's bitten
me before, when I was called on to port code (which I did not write)
from one machine to another. Having standard operators with
under-defined behavior is just inviting trouble: long debugging
sessions, or worse, unexplained failures in the field. Of course you
and I would avoid all the pitfalls at the start. :-)

.... and now back to your regularly scheduled Python newsgroup.

[Grant Edwards]

Re: negative integer division

On 2005-02-09, Jive Dadson <jdadson@yahoo. com> wrote:
[color=blue][color=green]
>> [C] isn't - it's a portable assembler.[/color]
>
> I've heard that many times, but it makes no sense to me.[/color]

I think the point is that C is a low-level, hardware twiddling
language to be used by people writing things like kernel code --
something that was always done in assembler before C came
along.
[color=blue][color=green]
>> Now, I'll agree with you if you want to argue that some
>> machines do negative integer division in stupifyingly horrible
>> ways.[/color]
>
> That's why I think it was a stupifyingly horrible decision.[/color]

For a language meant to write user-space applications where one
probably cares what happens when a division results in a
negative integer, it is a horrible decision. For a "portable
assembler" used to write device drivers it makes sense. People
writing that sort of code presumably know how their hardware
behaves, don't expect that everything write is portable, and
just don't do division with negative numbers. When they do
division, it's with postive numbers and they don't want to
waste the extra clock cycles to do it in a way that's
machine-independant for negative numbers.

The fact that C ended up in the rather inappropriate role of
a user-land application language is different problem.

--
Grant Edwards grante Yow! ... I think I'd
at better go back to my
visi.com DESK and toy with a few
commonMISAPPREH ENSIONS...

[Jive Dadson]

Re: negative integer division



Grant Edwards wrote:[color=blue]
>
> On 2005-02-09, Jive Dadson <jdadson@yahoo. com> wrote:
>[color=green][color=darkred]
> >> [C] isn't - it's a portable assembler.[/color]
> >
> > I've heard that many times, but it makes no sense to me.[/color]
>
> I think the point is that C is a low-level, hardware twiddling
> language to be used by people writing things like kernel code --[/color]

And Python interpreters?
[color=blue]
>
> The fact that C ended up in the rather inappropriate role of
> a user-land application language is different problem.[/color]

In the early 80's, either C was the "appropriat e language" or there was
none ... and that's coming from someone who wrote a commercial
Pascal compiler, runtime support, and debugger. I did it all in C.
Pascal, as we all know, was ill-conceived. C++ was a momentous
advance, but it intensionally inherited many of C's warts.

I've forgotten what we are arguing about, but I'm sure I'm right.

J.

[Jive Dadson]

Re: negative integer division

intentionally

[Grant Edwards]

Re: negative integer division

On 2005-02-09, Jive Dadson <jdadson@yahoo. com> wrote:
[color=blue][color=green][color=darkred]
>>>> [C] isn't - it's a portable assembler.
>>>
>>> I've heard that many times, but it makes no sense to me.[/color]
>>
>> I think the point is that C is a low-level, hardware twiddling
>> language to be used by people writing things like kernel code --[/color]
>
> And Python interpreters?[/color]

No. That's my point. C isn't a good language in which to write
user applications such as a Python language. One uses C when
the other choice is assembly, not when the other choice is
"real" high level language like Python, or Modula-3, or
Smalltalk, or whatever.
[color=blue][color=green]
>> The fact that C ended up in the rather inappropriate role of
>> a user-land application language is different problem.[/color]
>
> In the early 80's, either C was the "appropriat e language" or
> there was none ... and that's coming from someone who wrote a
> commercial Pascal compiler, runtime support, and debugger. I
> did it all in C. Pascal, as we all know, was ill-conceived.[/color]

I never thought so. I did embedded systems development using
Pascal and quite enjoyed it. Later when Pascal waned and C
waxed, I thought C was a definite step backwards. I thought
Modula-2 and Modula-3 were both good languages as well.
[color=blue]
> C++ was a momentous advance,[/color]

Now C++ _was_ ill-conceived. It's more of an agglomeration of
features than Perl. And dangerous to boot: at least Perl tries
to protect you from memory leaks and stray pointers.
[color=blue]
> but it intensionally inherited many of C's warts.[/color]

And added a bunch of it's own.

This is pretty much completely off-topic now. :)

--
Grant Edwards grante Yow! I'm definitely not
at in Omaha!
visi.com

[Grant Edwards]

Re: negative integer division

On 2005-02-09, Jive Dadson <jdadson@yahoo. com> wrote:
[color=blue]
> intentionally[/color]

I disagree!

--
Grant Edwards grante Yow! ... I don't like
at FRANK SINATRA or his
visi.com CHILDREN.

[Peter Hansen]

Re: negative integer division

Grant Edwards wrote:[color=blue]
> This is pretty much completely off-topic now. :)[/color]

No discussion of how lame other languages are is ever
completely off-topic in comp.lang.pytho n. After all,
these discussions continue to remind us how lucky we
all are to be able to program in Python, and that
can only be a good thing. ;-)

-Peter