On 7/9/11, Andrej Mitrovic <andrej.mitrovich@gmail.com> wrote:
>     double step = 1 / width;   // or .1

Woops that should have been `1.`. See, another bug right there!

On Fri, 08 Jul 2011 18:47:55 -0400, Andrej Mitrovic <andrej.mitrovich@gmail.com> wrote:

> On 7/9/11, Andrej Mitrovic <andrej.mitrovich@gmail.com> wrote:
>>     double step = 1 / width;   // or .1
>
> Woops that should have been `1.`. See, another bug right there!

1.0

1. is a horrible legacy thing, and should be removed.

-Steve

Andrej Mitrovic:

>     enum width = 100;
>     double step = 1 / width;
> ...
> This seems like a very easy mistake to make.

Right, it was present even in Python 2.x:
http://ideone.com/BEZqq
The bug you have found is common enough to deserve a so fundamental change in Python, they have improved it in Python3.x:
http://ideone.com/QeZEk

In Pascal/ObjectPascal you have both FP division (/) and integer division (div). In OCaML too, there is "/" and "/." to tell them apart. I believe such distinction reduces similar bugs a lot.

Even if Python2 was (and is) a very widespread language, and they have fixed the division, I don't think we'll see this change in D. Introducing an integer division operator in D is a possibility, but I think this is not enough to avoid your bug.

------------------------------

Steven Schveighoffer:

> 1. is a horrible legacy thing, and should be removed.

I agree, just as .1 :-) Saving one char of souece code is not worth the troubles this causes now and then.

Bye,
bearophile

On 7/9/11, Steven Schveighoffer <schveiguy@yahoo.com> wrote:
> On Fri, 08 Jul 2011 18:47:55 -0400, Andrej Mitrovic <andrej.mitrovich@gmail.com> wrote:
>
>> On 7/9/11, Andrej Mitrovic <andrej.mitrovich@gmail.com> wrote:
>>>     double step = 1 / width;   // or .1
>>
>> Woops that should have been `1.`. See, another bug right there!
>
> 1.0
>
> 1. is a horrible legacy thing, and should be removed.
>
> -Steve
>

I get tempted to use it because it's there and then get shot in the foot. :)

On 2011-07-09 01:03, Steven Schveighoffer wrote:
> On Fri, 08 Jul 2011 18:47:55 -0400, Andrej Mitrovic
> <andrej.mitrovich@gmail.com> wrote:
>
>> On 7/9/11, Andrej Mitrovic <andrej.mitrovich@gmail.com> wrote:
>>> double step = 1 / width; // or .1
>>
>> Woops that should have been `1.`. See, another bug right there!
>
> 1.0
>
> 1. is a horrible legacy thing, and should be removed.
>
> -Steve

Yes, please remove it.

-- 
/Jacob Carlborg

"Steven Schveighoffer" <schveiguy@yahoo.com> wrote in message news:op.vybo0hv1eav7ka@localhost.localdomain...
> 1. is a horrible legacy thing, and should be removed.
>
> -Steve

This I hate, but I'd hate to lose 1.f and 1.L

Steven Schveighoffer:

> 1. is a horrible legacy thing, and should be removed.

I have asked for this in two bug reports, both closed because of octal literals: http://d.puremagic.com/issues/show_bug.cgi?id=3837 http://d.puremagic.com/issues/show_bug.cgi?id=2656

So I have opened one more just for FP literals: http://d.puremagic.com/issues/show_bug.cgi?id=6277

Bye,
bearophile

On Sat, 09 Jul 2011 07:38:39 -0400, Daniel Murphy
<yebblies@nospamgmail.com> wrote:

> "Steven Schveighoffer" <schveiguy@yahoo.com> wrote in message
> news:op.vybo0hv1eav7ka@localhost.localdomain...
>> 1. is a horrible legacy thing, and should be removed.
>>
>> -Steve
>
> This I hate, but I'd hate to lose 1.f and 1.L
>
>

What's wrong with 1.0f and 1.0L ?

-Steve

"Steven Schveighoffer" <schveiguy@yahoo.com> wrote in message news:op.vygk71eweav7ka@localhost.localdomain...
>
> What's wrong with 1.0f and 1.0L ?
>
> -Steve

It's one character longer!

Mostly it's just a habit I've gotten into.  I'd likely be able to get used to writing the zero fairly easily.

On Mon, 11 Jul 2011 10:45:16 -0400, Daniel Murphy <yebblies@nospamgmail.com> wrote:

> "Steven Schveighoffer" <schveiguy@yahoo.com> wrote in message
> news:op.vygk71eweav7ka@localhost.localdomain...
>>
>> What's wrong with 1.0f and 1.0L ?
>>
>> -Steve
>
> It's one character longer!

1f is one character shorter :P  (BTW, not sure the grammar currently supports this one, even though it compiles. I think:

Integer FloatSuffix

should be added.

I admit, there is no equivalent for real, since L is dual-purposed.

-Steve