Re: [dev] Interesting post about X11

From: pmarin <pacogeek_AT_gmail.com>
Date: Mon, 21 Jun 2010 14:08:48 +0200

*should be used* and *can be used* have different meaning in my poor English.
Can you rationalize why Scheme *should be used *?

On Mon, Jun 21, 2010 at 7:09 AM, Robert Ransom <rransom.8774_AT_gmail.com> wrote:
> On Sun, 20 Jun 2010 16:00:44 +0200
> Alexander Teinum <ateinum_AT_gmail.com> wrote:
>
>> > Use Scheme.  See Scheme 48 <http://s48.org/> for a nice, simple
>> > implementation to start hacking on.
>
>> This thread is about a replacement for X, but we’re also discussing
>> development of “regular” applications. What exactly would you
>> recommend Scheme for?
>
> Scheme *should* be used for almost everything -- bootloaders, OS
> kernels, hardware drivers, tiny user utilities (like (Plan 9) ls and
> mc; Unix ls no longer qualifies as a tiny utility, and should not be
> written at all), long-running servers, etc. -- everything but x86 boot
> sectors should be written in Scheme.
>
> Unfortunately, the readily available Scheme systems are unsuited for
> most of those tasks.  At the moment, Scheme *can* be used for
> scripting and moderately large user applications (roughly, any daemon
> with a built-in or otherwise firmly attached GUI -- think mail UAs and
> multi-file editors for common examples).
>
> For low-level programming (kernels and drivers), you would need a
> Scheme compiler with support for compile-time and explicitly specified
> run-time memory allocation, as well as good type inference and support
> for explicitly specified physical types.  For small utilities, you
> would need a Scheme implementation with a small run-time library.
> Long-running servers would benefit from the same compiler capabilities
> that low-level programming requires, but you can usually do without
> them.
>
> Robert Ransom
>
Received on Mon Jun 21 2010 - 12:08:48 UTC

This archive was generated by hypermail 2.2.0 : Mon Jun 21 2010 - 12:12:02 UTC