[Date Prev][Date Next][Thread Prev][Thread Next]
[Date Index]
[Thread Index]
- Subject: Re: Lua-based OS (was: Safe Lua)
- From: Fabio Kaminski <fabiokaminski@...>
- Date: Wed, 3 Aug 2011 15:08:52 -0300
Particularly i think its a bad idea..
if its only to prove a point.. that you can create a Os in Lua.. then
it will be like that OS in haskell only..
just to tell its in haskell..
i think languages has fitness.. and the best program will be created
with the languages with the best fitness.. and of course.. the skilled
programmers..
and lua doesnt fit for low level.. it can be done of course.. but you
will end with a bad OS.. at least compared with a C or even a C++
based kernel..
if you want a proof of concept.. for the love of the language.. than fine..
But dont think the kernel will be anything competitive.. you will
need "byte-eaters" languages..
the other thing is if you wanna explore some OS experiments.. like
create a Exokernel based OS.. or a microkernel..
but this will be very scientific and academic..
Also the garbage collector will be "in the way" of a lot of
functionalities provided by OS, like memory management.. device
drivers .. etc
or you would have to fork Lua vm with a VM with manual memory
management.. or you would end writing in C anyway..
in the end nobody cares in what language does that OS was coded.. but
if its good and competitive..
But i think a good propose would be doing the core parts in languages
like C (where the fits for low level is high)
and where its possible.. for management.. shell... etc.. it could be
done in lua..
by the way Fabrice Bellard creator of QEMU, TCC, FFMPEG... create a PC
emulator written in javascript only
compiled a linux kernel and you can run it in your own browser in this link
http://bellard.org/jslinux/
anyway is possible.. but for lab experiments.. etc..
for usability and popularity i think its a bad idea.
On Wed, Aug 3, 2011 at 2:42 PM, Marc Balmer <marc@msys.ch> wrote:
> Am 03.08.11 19:36, schrieb Bogdan Marinescu:
>> Hi,
>>
>> On Wed, Aug 3, 2011 at 5:17 PM, Pierre Chapuis <catwell@archlinux.us> wrote:
>>
>>> On Wed, 3 Aug 2011 09:39:14 -0300, Elias Barrionovo wrote:
>>>
>>>> eLua ( http://www.eluaproject.net/ [2] ) was reported to run on a
>>>>
>>>> pentium with no OS underneath it:
>>>> http://wiki.eluaproject.net/**Booting%20eLua%20on%20a%20PC<http://wiki.eluaproject.net/Booting%20eLua%20on%20a%20PC>[3]
>>>>
>>>
>>> This is the kind of thing I was looking for, thanks.
>>> Sounds like I should take a look at eLua.
>>>
>>
>> Sorry to disappoint you, but eLua booting directly on a PC was merely a demo
>> which was meant to show eLua's high portability (also something I wrote
>> because I didn't have any eLua-capable micros back then). I didn't try this
>> specific port in a while and I'm not ever sure if it boots on the PC anymore
>> (it should); even if it does, you're going to have a really hard time
>> writing drivers for it. This is one of the main reasons that makes
>> implementing a new functional OS on a PC nowadays a quick and painful
>> failure. One simply can't keep up with all the hardware without a developer
>> base comparable with the one behind Linux. You could choose a limited number
>> of peripherals that you support (this is the approach taken by MenuetOS for
>> example), but then you'll end up configuring a PC for the sole purpose of
>> running your OS. Not many people are willing to do that. In the end such a
>> project is interesting, a lot of fun to work on, it has an extremely high
>> educational value, but don't expect it to be practical. For a PC, running
>> Lua on top of a Linux kernel sounds like a much better idea.
>
> Slightly related:
>
> It might of interest to some folke here that NetBSD contains Lua in base
> and that there is an (ongoing) project to Lua in the kernel as well. It
> was a GSoC project last year, run by Lourival Neta and mentored by me,
> and Lourival and I recently decided to again hack a bit on this.
>
> Prototyping device drivers in Lua, or do other "weird" stuff should be
> possible. Lua is loaded as a kernel module, but many, many things are
> yet to be decied on how to do it.
>
> It's an experiment....
>
>
>