BSD is dying

Paul Robinson paul at iconoplex.co.uk
Thu Jul 28 16:21:27 BST 2005


On Thu, Jul 28, 2005 at 03:25:25PM +0100, Paul Richards wrote:

> Actually I think they are. Unix hackers are ultra conservative when
> it comes to fundamentally changing the "unix way". If it was left
> to the choice of unix hackers then email would remain strictly text
> based.

This is not about HTML e-mail, but stay with me...

There are many arguments for keeping what is predominantly a textual
form of communication free of distraction. I should be able to decide
what my e-mail looks like, not you, because I know how I best absorb
e-mails.

If your e-mail requires spinning graphics and big red letters then
either:

1. It's better suited as a resource you direct me to like a webpage
2. It's spam, and I don't want it
3. You haven't thought through what it is you're communicating.

Now, I know lots of people will try and turn this into a HTML vs
plaintext e-mail issue. It's not. This is about the fact that the 'unix
way' is to think very carefully about function and process and implement
it. The resistance is not through conservatism - it's through
understanding.

> Would spotlight be a realistic possibility in FreeBSD. It requires
> hooking into the kernel filesystem API and indexing files as they're
> changed. That sort of digression from unix thinking is virtually
> unheard of in the traditional BSD environments. We have very poor
> sound support or gadget support, whereas Linux is very good and OS
> X is virtually flawless.

Yup, and I know that people would be interested in the discussion of
where that support would be hooked in. Should it be a fundamental part
of the filesystem? Should it be in the kernel and work regardless of
file systems? Would it not be best you show somebody how to use 'find'?
How do we go about extending metadata on the filesystem, and how do we
make that metadata searchable, and accessible? Do we dictate what
metadata should be added and therefore searched, or should it be open
tagging entirely left up to the user, with a couple of default schemas
they might find useful?

Think about all this. Write down some answers. Write some code. We'll
all say 'Wow, he's cool...' and buy you beer at conferences. People will
give you large salaries and offer you jobs in cool parts of the World.
People will want to sleep with you. Trust Me On This(tm).

> I hadn't intended this thread to be about desktop support, but I
> have to agree that Linux is eons ahead of FreeBSD and that is because
> of shortcoming in the OS. Things like a gui based installer, better
> packaging and automated binary updates.

On a side note, I think most people who cared about installers, came to
the conclusion several years ago that 'GUI' is not the problem.
Customisation and scalability of the installer is the problem, and you
should be able to 'skin' it with a GUI if that's what you want. However,
just turning /stand/sysinstall into a GUI version is *not* what is
needed.

Better packaging though, oh yesyesyesyes please. Decent binary updates
that is supported out of the box, yes, we want that. Being able to
migrate through version numbers without having to mergemaster and
buildworld, or having to reboot with a CD in the drive (that is 3,000
miles away), yes, that would all be good...

This thread today has reminded me why my outbox used to average 1.2
million words a year a few years back... :-)

-- 
Paul Robinson




More information about the Ukfreebsd mailing list