Re: Thoughts on a New Dylan Ecosystem

What about adopting something like GMP[1] and doing something similar to
Common Lisps bignum/fixnum dichotomy where you use the native integer size
where appropriate, but get arbitrarily large integers outside the range?

    -tree

[1] http://gmplib.org/


On Wed, Nov 13, 2013 at 6:22 AM, Bruce Mitchener
<[email protected]>wrote:

> Carl has brought up a couple of issues to me privately that I hadn't
> addressed here.  I'll leave some for him to discuss, but one really
> important one is our 64 bit story.
>
> This ties back into general issues with our numerics as well.
>
> Running Dylan on a 32 bit platform is pretty limiting due to the range of
> <integer>.
>
> Notably, I have these issues with it:
>
>
>    - Timing operations in nanoseconds are difficult.
>    - File routines involving seeking within a file.
>    - Simple Random's <random> support is for <integer> only, so highly
>    limited.
>
> It isn't a great solution to have to use <double-integer> as that is
> somewhat awkward, expensive, and isn't fully functional.
>
> Having a much improved story in this area is very important.
>
>  - Bruce
>
>
>
>
> On Tue, Nov 12, 2013 at 5:07 PM, Bruce Mitchener <
> [email protected]> wrote:
>
>> I've written a blog post:
>>
>>     http://dylanfoundry.org/2013/11/12/thoughts-on-a-new-dylan-ecosystem/
>>
>> I've posted it to Reddit:
>>
>>
>> http://www.reddit.com/r/programming/comments/1qg3uw/thoughts_on_a_new_dylan_ecosystem/
>>
>> and on Lobste.rs:
>>
>>    https://lobste.rs/s/nj6i27/thoughts_on_a_new_dylan_ecosystem
>>
>> Discussion, comments, etc are welcome!
>>
>>  - Bruce
>>
>>
>
> _______________________________________________
> hackers mailing list
> [email protected]
> https://lists.opendylan.org/mailman/listinfo/hackers
>
>

_______________________________________________
hackers mailing list
[email protected]
https://lists.opendylan.org/mailman/listinfo/hackers