Re: [Mingw-cross-env-list] Re: Mingw-w64 thoughts (was OSX 10.6 workarounds)

[Tony Theodore]

2009/11/15 Volker Grabsch <address@hidden>:
> Tony Theodore <address@hidden> schrieb:
>> Is there much interest in win64 applications at the moment?
>> Your thoughts.
>
> While my answer is technically "Yes", there are political
> and moral issues I'd like to discuss with the whole group
> of Mingw-cross-env users and developers. To the point, the
> main question is:
>
>    Should we really do that?
>
> Since the beginning of this project I had mixed opinions
> about Windows porting. On the one hand, you increase your
> userbase this way, and you help users who must use Windows
> for some reason. On the other hand, you're putting much
> work into Windows portability which essentially just helps
> a monopolist.
>
> As a starting point, I ask everyone to skim through the
> following documents before engaging in the discussion.
> They provide views from many different directions and
> cover almost all important arguments:
>
>    http://www.gnu.org/philosophy/microsoft.html
>    http://www.fefe.de/nowindows/
>    http://catb.org/~esr/writings/world-domination/world-domination-201.html
>    http://kde-cygwin.sourceforge.net/faq/main.php#9
>
>    (Please complement this list if you feel there's an
>     important document missing here.)
>
> My current answer to this moral dilemma is:
>
>  1) I trust you, the developers who use Mingw-cross-env, to make
>     the right strategic decision for your projects. Mingw-cross-env
>     ports just libraries. It's you who port the actual applications.
>
>  2) I remove any library whose author doesn't want it to be ported
>     to Windows, even though the license wouldn't forbit porting:
>     http://hg.savannah.gnu.org/hgweb/mingw-cross-env/rev/35b9dcbe9daf
>
>  3) At least I help _developers_ to get away from Windows. Helping
>     _users_ to get away from Windows, if they want, is a different
>     battle.
>
> However, the "World Domination 201" article suggests that this
> might apply only to win32, not win64.
>
> In the past, I wasn't able to discuss that issue with a big
> audience. But luckily, the Mingw-cross-env project grew, so
> now I do have an great audience to ask:
>
> What do you think about porting software to Windows in general?
> What do you think about supporting win64 in addition to win32?
>
> A penny for your thoughts.

Fascinating reading, I hadn't thought about such things for a long time.

Mostly these days I think about windows with a mix of ambivalence and
pity. I see people struggling and being hampered by technology -
rather than using it to enhance and enrich their lives. It's actually
the proprietary ecosystem that's the main problem, people are trapped,
and every piece of ported software is a step on the long road to
freedom.

Simply having alternatives though, is not enough. Apple are so closed
that it's almost impressive (most governments can't keep state secrets
as effectively). The greater issue is challenging people's notions of
intellectual property and the way software is developed. The larger
projects (say firefox) don't serve to make people stay on windows
longer (they're not moving anyway) - it's an entry point to their own
exploration of free software (for those who care) and base of
discussions for learning a new paradigm.

I (we?) take free software (and software freedom) for granted, but
it's still alien to most people. It's easy to forget that people don't
choose windows -  they have it. Porting software benefits these users
directly with it's immediate utility and indirectly with the reduced
reliance on a proprietary ecosystem. It's this exposure and gradual
mind-shift that will ultimately cause these users to generate demand
for an entirely open operating system.

At least, that was my experience. Back in 2000, I was using Interbase
(on AIX), when it was open-sourced. I didn't know what that meant,
other than the new FirebirdSQL had windows installers - for free!.
That started me on a path to the wider world of free software, but it
was still about two years till I tried my first linux install. If
there hadn't been windows versions of these things, I never would have
been compelled to try a different OS - and if I had to wait for a
pre-installed linux... Now I wish someone had shown me earlier.

So I see porting in much the same light as i18n or translation of
literature - a worthy pursuit that I can fully support. Some things
are lost, but it's the underlying concepts that are important, the
technical barriers are enough on their own - no need to insist on
conformance to language/platform.

I don't see this changing in the 64 bit transition, the principles of
exposure and exploration remain - however few pursue it. The 64 bit
was/is a great opportunity, particularly for linux vendors who could
win a "driver war" if they pitched it that way - "Plug it in and it
works". What I'm seeing is a hardware related shift in the other way.
People are moving to smaller, more portable devices. There's Maemo,
Andriod, and the most tightly controlled system I've ever seen - the
iPhone OS.

Where does that leave mingw-cross-env? Desktop systems aren't going
away, and 64 bit is the future. In my mind, it's applications that
actually drive hardware adoption - it's hard to use 4GB without
db/filesystem cache, multi-media, or virtual machines. I don't think
2008 was a hard deadline, and windows 7 has given everyone another
pass. It has also upped the ante. There's a Subsytem for Unix
Applications (SUA) which is a revitalised Interix/Services For
Unix(SFU). I can find very little information about it, but it's the
only compelling reason for me to look at win7 - yet not so compelling
since nearly a year has passed since I heard about it.

So I think that supporting win64 in addition to win32 is a necessary
step in the evolution of the project.

Then again, I'm not tied to any particular operating system. I see
them as a substrate with their own low level features that
applications can utilise or not. I'd like to use zfs and dtrace on
linux, oracle on freebsd, pf everywhere. I'd probably like a
cross-compiler that gave me something like freebsd ports on any OS. If
I could do

cd /usr/ports/pkg-to-build
make pkg --arch=x86,x86_64 --os=freebsd,lin_rpm,lin_deb,osx,solaris,win...

I'd be happy.

Tony