[Gslug-general] The next generation of Linux (or what it should be)

[Andrew Beyer]

On Tue, Sep 1, 2009 at 12:13 PM, Joe moo <starquestnerd at gmail.com> wrote:
> One of the reasons Linux is not as popular as it should be in my
> opinion is inconsistencies. It is good to have many different
> options(distros) especially for Power Users (like myself) and
> Developers. However, the problem with that is all of the
> inconsistencies between them, this is a big issue for end users and is

I won't argue that it isn't an issue, but I fail to see how you could
resolve all the inconsistencies without sacrificing  most of the
benefits of having the variety in the first place. There are things
that _should_ be consistent, where consistency does not meaningfully
impact people's choices,  but most of them already are...POSIX, LSB,
freedesktop.org, and others have already resolved most of the
low-hanging fruit here. I think the things that they haven't
approached or haven't succeeded in standardizing are mostly those
where by mandating a standard you take away choices people want to be
able to make.

> probably why the end-user market doesn't extend much farther than
> netbooks.

Major hardware vendors marketed netbooks with linux to a general
audience. That hasn't happened on the same scale for any other class
of hardware (at least any comparable to a pc -- I'll ignore more
deeply embedded devices for the moment), so I don't think its really a
valid comparison. The vast majority of people use the os that ships on
their computer and are unlikely to switch...many don't know they can.

> I believe that it is time to set some standards for all of
> the projects out there; standards that can be agreed upon and followed
> by the majority of projects.

Lots of people have tried...good luck.

> is because Windoze has an architecture
> that most applications follow

I'm going a bit off-topic here, but I'd take issue with that. For a
single entity, Microsoft can put out an amazing and bewildering array
of related and overlapping technologies with the attending
inconsistencies and incompatibilities. The core win32 api is
relatively clean and consistent, but is pretty much comparable to what
posix offers...not enough of a standard to build on from scratch. I
think .NET alleviated, if still not resolved the issue. But
particularly before then, writing a significant application on windows
involved a fair bit of black magic making Win32, MFC, ATL, COM, COM+,
ActiveX, the VB runtime, et. al. cooperate unless any one of them
provided everything you needed, and you didn't interact with any other
software that used any of the others.

>  WX is one of the greatest examples of a simple cross
> platform architecture
> ...
> One solution to that is the
> pitifully unused but amazing autopackage framework!

Both technologies which, even if you can argue their technical
superiority to other options, are already uncommon/underrepresented.
Try to build a standard on that, and you get a "standard" on paper
which no one conforms to. Adoption has to come before standardization
unless you just want to put out well intentioned documents, or can
force people to comply.

If you really think these are the solution, your task isn't
standardization, it's advocacy... you need to convince people to adopt
them. Get a modern and full-featured web browser and office suite to
adopt WX as their gui platform. (if you suggest this on the mozilla or
oo.org lists, please don't cc me on the resulting flamefest :) Get
everything (or some significant subset) in Ubuntu's universe or
multiverse (or some other sufficiently large set of software) packaged
as an autopackage file, and publish a central collection somewhere.

> Linux needs to stick to a standard that users can follow if it is to thrive!

Does it? Why should every distribution that happens to use the same
kernel also make all the same choices and trade-offs in userspace? If
you want a consistent set of choices already pre-made  for you, why
not just stick with windows? Or choose a single linux-based distro and
use the software packaged for it? If you want to make your own set of
choices, what makes you think they will be any more generally
appropriate for everyone else than the ones various distros have made?

> PS. No I'm not putting down Linux I'm just suggesting new ideas!

And I'm not putting down your ideas, but there have never been a
dearth of people with ideas of how linux should "standardize",
particularly when said standard would include their favorite bits of
software. I'm not saying that there aren't huge improvements to be
made in many areas, but I think the solution is to develop and/or
spread the better software, not to rhapsodize on the merits of it
being declared a standard.