h a l f b a k e r yAlmost as great as sliced bread.
add, search, annotate, link, view, overview, recent, by name, best, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
Please log in.
Before you can vote, you need to register.
Please log in or create an account.
|
Autodistro
A witch which asks you which kind of linux distro you wish. | |
There are many linux distributions. Many are similar to each
other,
and the rest share the characteristic that they are not as similar to
each other. Ultimately, all linux distributions are linux,
underneath
it all (not very far underneath, either).
My idea is for a web application that
acts as a witch, which guides
your answers by asking questions which will prompt said answers.
The result becomes a 'profile' and will be compared with existing
profiles, for ease of management.
A profile, then, might supplant what we know of as a distro. It is
then used to 'assemble' and 'deliver' your specific distribution of
desire.
Rather than choosing to install, eg, Debian, or Gentoo, or Fedora,
or
Suse, or Ubuntu, I might have tried one or two of these already
and
so I've been exposed to things about them that I like. For
example,
the package management system and depositories (eg: rpm, debs,
or
compiling things from source with your optimised make.conf such
as
Gentoo's portage uses). Or the variety of processor types it can be
optimised for (again, see Gentoo). Or the speed of updating and
installing stuff (avert your gaze from Gentoo). Or various other
aspects that might be appealing to you.
To start with, each of the established distros would have profiles
on
the Autodistro service - for example, Xubuntu 8.10 (intrepid)
would
be downloadable from the usual ubuntu mirrors, but also
synthesizable from Autodistro, which would create a more or less
identical iso or other delivery method using the appropriate
profile.
The option might also be viable to request a very old distro if the
profile were rigged up, and the sources were still available, but
I'm
not sure there's much appeal in retro software. But still, it could
be
done.
Where it might be of disruption would be to abstract out what
these
exact differences are in terms of aspects that can be specified,
rather than aspects that are attributes of the distro you happen to
be
using. For example, it might be viable to, via the question-asking
witch, to select an Ubuntu-netbook-remix -like distro, except that
the package management should be portage, and the WM should
be
xfce. For example. That gets me as far as the kind of differences
I'd
be faced with in choosing a whole distro, or sub-flavour of one.
One
final question at this level might be "livecd; liveusbstick image;
distro installer; or combinations?".
At a further level of tailoring, the witch would be present more
detailed questions, such as which languages can you do without,
keyboard layouts, etc. Also, what the target machine is (eg, an
Acer
Aspire One netbook) and all the appropriate drivers and such are
included - or alternatively, keep it general, and your distro will
work
on all machines - useful if you selected liveusbstick or livecd.
Then, at a much more refined and personalised level of mass
customisation, you can pick 'families' of software for 'in' or 'out'.
Of
course, this is just your start image, and of course, you can
proceed
to use or install the distro and install any further software you like
in
the usual manner. It's just that with many livecds I use, there's a
predictable bunch of software that tends to be there and I'll never
touch it, so I'd simply select those groups out altogether (eg,
games).
A more typical use of this level might be that I'd experiment with
an
Autodistro profile that gives me a liveusbstick netbook image that
only contains a browser - no games, no office software, no
graphics
software, no media players. Just to see how it feels. At the
moment,
I'm achieving that by starting with a 'proper' livecd or liveusbstick
and then subtracting the stuff I don't want. I'd rather start that
way
and build up later.
Finally, your Autodistro profile will be compared quantitatively
and
qualitatively with others, and 'very similar' ones will be made
aware
to you (and others, similarly) and you might, for the sake of
efficiency, elect to use an existing known one that highly
resembles
the one you just specified, as a base from which to tweak.
Therefore
the Autodistro application could grow to form some sort of
relational
inference regarding what is 'similar' or not, from a user
perspective.
Ultimately, with intelligent use of profiles and a large profile
base,
it might be possible for a completely new and novice user to be
guided towards a very usable distro for them, without their
needing
to know much at all about what they're about to be doing.
Hopefully, it will result in easy to use and appropriate
distributions
that 'just work' for the average user, and highly customised
distribution profiles for the expert twiddler, also.
searchme: with halfbakery as the query
http://www.searchme.com/#/&q=halfbakery/
I am in no way affiliated with the product... [4whom, Jan 01 2009]
[link]
|
| |
Or... compile your own kernel. Like any novice wants to do *that*! (If I had a penny for every poorly made kernel I'd come up with) |
|
| |
Ubuntu seems like it knows what it's doing when it installs but it certainly isn't as lucid as the Autodistro would be useful. |
|
| |
Yes - sounds like a reasonable idea |
|
| |
Would this be like an automated Linux nerd (flying on a broom) who is familiar with at least 20 distros in their latest form, and comes to your house and interviews you, then chooses the best for you? Sounds like it would be an enormous amount of work to keep current. I like linux so[+]. |
|
| |
//a web application that acts as a witch//
Providing eye of newt, etc? + |
|
| |
Aren't they all much of a muchness ? Redhat, 10 years ago had the family of stuff chooser etc etc. Last time I checked Linux, you only realised you made the wrong choice when you changed your mind post installation. So I guess this does fall more into the "Geek on a broom" distribution rather than an "Autodistro". Unless all the various package managers play nice, but that would rather make the different distributions redundant. |
|
| |
Probably of more use is a web system that offers a "a second take" profiling where a distribution is suggested, but you can then see the profiles various users moved on to. Or in other words "9 out of 10 visitors found our service useful. The remaining 10% knew what they were doing and didn't come back. The best you can hope to get from this service is the judgement of a novice." |
|
| |
It is a good example of evolving decision trees and how the internet fails to deliver the spread of information in any qualified context. Like the dilemma about the "forest tree voting" idea, this idea demands "decision tree credibility". In essence we do not have an information architecture where you can distinguish between a novice and an expert beside the place you saw the information or if you keep up with usernames, who wrote it e.g. "Joel is a bit green but ok" or "Wikipedia is ok, not too much graffiti". |
|
| |
Who's maintaining this web site? To whom do its customers turn for support? |
|
| |
I was thinking of support while I was writing it, but didn't
know whether this overlaps the commercial or professional
support+maintenance
product providers or not. I suspect, probably not. This
degree of variance would be support hell. |
|
| |
On the other
hand, if fans of linux drift from allegiance to various
distros to allegiance to Autodistro, then you'd find people
who inherently give good support (and there's a goodly
supply of those heroes on freenode, for example) would
find themselves in a wider nodal pool. |
|
| |
For example, an ex-Mandrake expert who now finds
themselves an old hand at
urpmi and consequently is a valued asset in Mandriva
support, would now also be exposed to the sort of bustle
that the .deb masters at Ubuntu are faced with, and both
would now overlap with the Red Hatters that help the
Fedora people out, and all of those would stand agape and
aghast and akimbo with
the <hyphenated>compile absolutely everything from
source</> crowd that
is the Gentoo audience (then compile it again because by
the time you've finished compiling it a new version of the
app has been released). All of this would, of course, inherently by now be one big mixup mashup of flavours
and architectural choices, rather than discrete distros. |
|
| |
I consider it another set of maze
wall boundaries that'd be hedge-trimmed down for the
people to walk over with no consideration. |
|
| |
I think you're overestimating (or misrepresenting) your audience. Evidently they know they want to run linux, but not which flavor. They know what features they want, but can't figure out which distributions offer them. |
|
| |
I'd be less worried about support (you could just throw your hands up and say "not my problem") than having a business case. I'd expect it to take a great deal of effort not only to stay up-to-date on all the distributions and their features, but to make sure your witch/wizard stays up-to-date as well. That's going to be expensive. |
|
| |
Yes. Although, arguably, linux support is presently balkanised
in that there are 'freelance' (ie, on freenode) experts in this
distro or that distro but rarely completely agnostic experts
that have equal exposure to all distros (after all, we all have
our favourites, and we all have a bunch that we've never
tried). The commercial providers of support and
maintenance (ie, the 'paid-for' versions of the distros) are
entirely walled into their own product, more often than not. |
|
| |
Wasn't there a music project with a similar idea, you listed artists you liked, and the engine would suggest new artists you might like based on similarities? If a user could try say, 3 specific distros, then rate aspects of them, then the engine could recommend a distro that suits the user. That might not be to hard to maintain. It would also filter those unwilling to at least put the effort into trying the 3 distros. |
|
| |
I think you're thinking of Pandora radio, am I right? |
|
| |
It always comes down to hardware interrogation. Software always runs sweetly with the knowing drivers. |
|
| |
//it's a choice of bewildering flavours and complex textures// |
|
| |
Linux cook-off. Now thats an idea for a show. |
|
| |
<large whiny post deleted> |
|
| |
It's a good concept, but with the (lack of) state of "AI" for help-software, it won't fly. |
|
| |
For instance, I neither need nor want User Accounts or ACL's, etc. in my distro: it's a single user (or occasionally supervised guest) machine so to me it's just bloatware; Unix was written for a different environment. |
|
| |
What distro/subset can do that for me ? |
|
| |
(-) I think I feel similar about this as phoenix does, sorry - this is adding complexity that software won't be able to sustain to an area already suffering from too much of it. Investing the same amount into just making the base set clearer/work at all would be more useful. |
|
| |
FlyingToaster, - Linpus Lite. |
|
| |
jutta, - the trouble with such efforts as the Linux Standard Base (LSB) and other 'standards' is that developments and innovations* quite quickly outpace any standard, bypassing it or rendering both the presence of a standardised way of doing something, and a novel but advantageous way of doing something, a mess. |
|
| |
* I never thought I'd refer to linux as innovative, but I think within the past two years it's now turned the corner from an arena of copycatting to a hotbed of true innovation. |
|
| |
§ I also thought I'd never come to see negative value in standards, but the more I think about it, the more I see standardisation as a hopeless inclination toward technological fundamentalism. |
|
| |
The last few annos really spell this out for me. A lot of people just want the most stable system and for that any distro that offers the features required is probably ok. The problem comes when you can't wait for a particular innovation because its still new and maybe restricted to a particular distro. Whether you are developing or just using, being part of a 'bleeding edge' area is about the same - both uses are taking part in innovation. So I guess the real purpose of this idea is to suggest the distro that is currently active in the area(s) of innovation you wish to participate in ? (For my money PVR's) |
|
| |
[Ian] since a quickly googled excerpt from their manual reads "At default settings, the password for the user account is “linpus.”" I'm thinking not. |
|
| |
Well, on my Acer Aspire One, which has Acer's mix of Linpus Lite, it doesn't have any user accounts (other than one called 'user' - but the normal people don't see that) - it just starts up and you're at the bit where you choose whether to browse the interweb or play games or do other stuff. The idea that there are actually such things as user accounts may well never occur to a typical AA1 user. |
|
| |
Lots of important software requires a multi-user system which is why you don't see single-user systems so much. |
|
| |
My iPod Touch has a single user - at least from the point of view of me (the single user). Obviously, as it's a unix system behind it all, it no doubt has the usual raft of other 'users' that aren't people. |
|
| |
Standardisation is sooo last year. Consumers, marketers and manufacturers are all moving to "personalisation". The trick is getting these very large combinations and permutations succinctly described and accurately recorded. It is the future, and it will come. |
|
| |
Eventually several users, even on one device, will have access to their own preferences as well as best practice preferences by users with similar profiles. Although there is a propensity for privacy intrusion, it does not have to be the case. |
|
| |
<gentle aside> Has anyone tried searchme:. A visual search engine with stacks (for personalisation). A bit like onenote meets a web browser. Will post a link. <\ga> If content can be moulded to a user why can't an OS? |
|
| |
Not sure about OS distribution, but I recently purchased a computer with MS Office product set pre-installed. From what I can make out, the key I input decides which subset of the pre-loaded product set I get. |
|
| |
I cannot see how this idea is different enough to garner such negative (well, less positive) response. Sure, the combinations make it jump off the charts, but not impossible. |
|
| |
Which begs the question. Why don't MS let you choose which products of the set you would like to activate for a certain fee? I would have loved the new Access, but couldn't give a toss for the new Word. Presuming they are both preloaded and awaiting activation. |
|
| |
my apology Ian, no intention side rail your ideas'
discussion . honestly . |
|
| |
\\ <gentle aside> Has anyone tried searchme:. A
visual search engine with stacks (for
personalisation). A bit like onenote meets a web
browser. Will post a link. <\ga> \\ |
|
| |
cool, as in colliris . my little umpc handled it
along with my voice recognition and onenote and
chrome, firefox, seamonky. yahoo im ( voice n
video going) and logmein and a music player
running ! |
|
| |
[IT], I think [FT] wants something that is literally single user, for whatever reason. |
|
| |
Then boot linux into 'single user' mode. There's a googlefull of pages out there in response to "single user linux" - doesn't take a special linux to do it with. Or, boot OS X into single user mode (hold command+s down when booting). |
|
| |