[GEM Development] GEM on Atari

[Liam Proven]

2009/1/6 Alan Hourihane <alanh at fairlite.co.uk>:
> On Tue, 2009-01-06 at 14:47 +0000, Liam Proven wrote:
>> 2009/1/5 Shane Gough <goughsw at gmail.com>:
>> > Hi Alan,
>> >
>> > When you say 'Gentoo running on FreeMINT' what do you mean? I would
>> > have thought that Gentoo (being a Linux based OS) would run directly
>>
>> I think Alan is getting confused, and thus confusing the rest of us.
>
> No, I'm not getting confused, and certainly not trying to confuse the
> rest of you.
>
>> Gentoo, as you say, is a Linux distribution: a complete OS based on
>> the Linux kernel.
>
> Gentoo/Alt
>
> http://www.gentoo.org/proj/en/gentoo-alt/index.xml

But you didn't say Gentoo/Alt or Gentoo-Alt. You said just Gentoo!

>> You can't run that on MiNT. It's not possible in any useful way.
>>
>> However, one of Gentoo's distinctive features is Portage, a
>> source-based packaging system, derived in inspiration at least from
>> the FreeBSD Ports system.
>
> Right, which is what Gentoo/Alt is all about.
>
>> (The other BSDs - NetBSD and OpenBSD - use something similar. I don't
>> know about DragonflyBSD and PC-BSD has its own, different binary
>> packing system, as well as Ports.)
>>
>> It may be that someone is trying to implement Portage on Mint. That
>> would be doable, although when I imagine the MiNT-based world is much
>> slower-moving than that of x86 Free software, I don't see a lot of
>> point.
>
> I've ported Portage to FreeMiNT and it's working very well here.

Good for you!

>> Secondarily, the point of Portage and of Gentoo in general is to
>> compile from source on every machine. There are 2 reasons for this,
>> neither of which apply on Atari & compatible hardware:
>>
>> [1] x86 PC hardware is fast enough that recompilation is quick
>> [2] There is such a diversity of PC hardware - CPUs alone in a dozen
>> families or more - that significant performance improvements can be
>> got from compiling with specific optimisations for each individual
>> machine. E.g., on Intel alone, processors include
>> 386dx/386sx/486slc/486dx/486dx2/486dx4/Pentium 1/MMX/Pro/2/3/4/4HT/4
>> 64-bit/4 64-bit HT/Core/Core2/Core2Quad/Atom/Core i7. That's 20
>> variants and I've not included 64-bit variants of everything since
>> Core2. Add in buses, caches, chipsets, graphics chips, choice of Linux
>> desktops, choice of version of GCC, choice of kernel, etc., and there
>> are millions of permutations of PC.
>>
>> How many 68030-based ST-family machines are there? Half a dozen?
>
> There is a 68060, 68040, 68030, 68020, 68000 for Atari architectures,
> and that alone is reason enough for me to deal with.

Doesn't MiNT require an MMU?

And if not - go on, how long to compile a full system on a 68000? 8¬)

> Secondly, the advent of cross-compilers makes it much more viable too.

Does Portage help with that?

>> So, with a small library of software, relatively slowly changing, on a
>> small number of fairly fairly homogenous machines, where available
>> processing power is so low that compilation will take hours, there
>> really is no point in Portage that I can see.
>
> Maybe from your point of view.

Well, go on then... How much active development is happening in the
MiNT world? Is any ST-family machine powerful enough to run GNOME or
KDE, say?

>> A free, open, GPL, binary-based packing system with automatic
>> dependency resolution built in from day 1 at a deep level, though,
>> would be a boon. That means APT, which is all that, and runs on both
>> .deb packages with DPKG and .rpm packages with RPM.
>
> Another, from your point of view.

Well, yes, but then, I evaluate and review operating systems and
tools, both for publication in about 20 different professional
magazines, journals and websites, from the Inquirer to Personal
Computer World and PC Magazine, and I recommend, build, install,
support and train users on systems for hundreds of companies from
5-man small businesses to multinational banks and financial
institutions; the biggest system I've supported so far has 38,000
users.

So, you know, I'm not just an amateur fiddler. :¬)

And I've tried Gentoo and my conclusion was, basically, that it was a
toy for "ricers". (Kids who customise their kit for no real purpose
except benchmarking competitions.)

Sorry. No offence meant, and I am not making any kind of personal
comment, judgement, accusation, or anything of the kind against you.

On the other hand, my considered professional opinion of Gentoo is
that it wasn't worth the time or effort and that Portage was clever
but really didn't offer anything genuinely useful for a home or
professional Linux user on x86 hardware.

For home users, in my opinion, Ubuntu is about the best there is.
Fedora is a waste of time, unless you like playing with alpha-grade
testing systems, and there's really no argument for Mandriva any more.
Ulteo is starting to look promising, maybe.

For enterprises, there's a case to be made for RHEL (possibly
supplemented with CentOS) or SLES. Debian for the pros, Xandros is a
viable corporate desktop, and frankly, that's about it.

But if you're compiling FOSS apps for different models of ST, yes, I
can see that Portage would be useful. The rest of the Gentoo userland,
though, I would think would be *way* too heavyweight for an ST, even a
maxed-out TT or Falcon, or even a Medusa or Milan - no?

-- 
Liam Proven • Profile: http://www.linkedin.com/in/liamproven
Email: lproven at cix.co.uk • GMail/GoogleTalk/Orkut: lproven at gmail.com
Tel: +44 20-8685-0498 • Cell: +44 7939-087884 • Fax: + 44 870-9151419
AOL/AIM/iChat: liamproven at aol.com • MSN/Messenger: lproven at hotmail.com
Yahoo: liamproven at yahoo.co.uk • Skype: liamproven • ICQ: 73187508