Slashdot Mirror
The Linux Newbie Replies: WFM?
Thanks to Sensei^ for sending us a
link to a piece about dealing with new Linux users. Given the gigantic growth of Linux over the last year, this is an issue within the community: How do we deal with this influx of population? Everyone recognizes the need for more documentation, but what's the best way to do it? If you've got an opinion about the whole schmeal, click below and add to the conversation.
Two years later I delved into Linux and have been using it for a year and a half.
Although Linux has an extensive set of documentation and I wasn't a novice to Unix by then, I still found FreeBSD easier to learn even though I was virtually new to Unix.
I found that the documentation for FreeBSD was organised very much like a manual which was great to start with as I could work through it as I installed a system. The topics were also general such as 'disks', 'backups', 'serial communications' and it was easy to quickly find what you needed.
The Linux docs are organised by a specific 'need' and while it is great when you are after a solution to a specific problem, it is too unstructured for a newbie.
A newbie needs to be lead through the topics in a general fashion so that they can gain an overall picture - Linux docs don't really give a good overview.
It's 5am so I'm not very coherent at the moment - so i apologise if my opinion isn't very clear. I can't really give specifics (at least I can't think of specifics at the moment), but that is my general feeling about what makes Linux that little bit more difficult for newbies.
...if every Linux user dropped the attitude and helped the newbies instead of getting off on how superior they are because they know Linux.
Simply reading the manual/readme/whatever doesn't always work. If someone is used to OS xyz or 123 jumping to Linux takes a little help. I think that the Linux community doesn't want average Joe to use Linux because that would put average Joe on the same level.
"If you do not believe in the freedom of speach for those who you despise, you do not believe in it at all." -Noam Chomsky
I find it hard to hear such words springing forth from my own mouth, but the concept of a "solutions database" seems remarkably appropriate here.
Generally, I rail like nothing else against solutions databases, because my job is to enable people to solve complex problems in complex systems, and solutions databases tend to grossly oversimplify this task, which I find mildly insulting. also, we must consider the 90%/10% rule, whereby 90% of the calls/problem reports encompass 10% of the possible problem domain, but, conversely, the 10% of the calls which involve the other 90% of possible problems will also consume 90% of support staff's time.
However, we are talking about newbies here. And newbies happen to consitute 99% of the 90% easy calls. Thus a solutions database becomes a realistic and productive answer.
blah blah blah solutions database blah blah solutions database blah.... okay, WTF do I mean by solutions db?
think search engine, but with better semantics, and an astonishingly high signal to noise ratio. the actual database consists of answers -- and only answers, so we don't frustrate users by matching their query with an unanswered question -- which are concise, correct, specific, cross-referenced to more detailed information (THERE'S TFM). these answers will be submitted by whomever (one thing linux has is an amazing volunteer support base), but (here's the catch) must be filtered, combed, possibly edited, by a dedicated group of individuals.
What's the value add beyond the many linux-based search engines, specific help pages, newsgroups/email reflectors, etc? specificity and conciseness -- in short, a very good signal-to-noise ratio. the key is not treating this like a damn swiss army knife -- build one tool that serves one purpose well, rather than doing a shitsplatter job trying to cover everything with one tool.
if anyone is interested in hosting such a site, I'd be glad to help make such a thing a reality.
nathan
*urp!*
(Ok, so I lied about it being 10... :)
It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
The fundamental problem right now is that no one knows what format their documentation should be in, and many people just punt or write minimal documentation as a result. HTML is useless for creating any kind of structured searching (unless you layer a documentation standard on top of it). roff has a decent documentation standard for UNIX, but no one likes it any more. texinfo is nearly impossible to manage since it requires sophisticated tools that don't play nice with anything that anyone actually uses. Plus, it requires a central table of contents which is difficult to manage automatically from un/installation scripts.
I've been thinking about it, and I really believe that Larry Wall's greatest contribution to the world has been POD (not Perl, itself). POD is a very simple documentation format that can be used to follow the roff-ish manual conventions of UNIX, but the format is so simple that it can be converted to man, HTML, texinfo (though texinfo standards usually want more prose than a UNIX manpage has), plain text, etc. This is a very nice solution for someone who's ambitious enough to go through the entire Linux documentation base (HOWTOs, FAQs, man pages, texinfo, PostScript, etc) and convert it all to this one format. Then each distribution could choose it's pet format to render in (probably *both* HTML for the new people and man for those who have "man -k" hardwired into their brains).
It would be nice to layer a few additional features on top of POD:
T which takes a term or phrase and indicates that this particular section of this document defines that term in a way that should be indexed globally. This is not quite the same as LaTeX's indexing. More of an HTML "A NAME=" sort of thing, but where HTMLs mechanism could be called a pull model, T would be a push.
H which takes a semi-colon separated URL and filename. The filename is an image that should be used as a figure in the document (numbered from one on). If the URL is provided, it is the location that should be referenced when users view this document in a text-only setting. The lack of images in POD is the only thing I don't like, and I know that it's quite unreasonable to expect that all Linux users (or UNIX users in general, for that matter) will be viewing documentation under a windowing system, but it would be nice to be able to show diagrams and other figures when the possibility exists.
Given these minor changes, rewriting the documentation would consist of converting all of the extant documentation over to text and then hand-hacking it back to POD (POD is very nearly plain text, with minimal markup that makes HTML look like a general purpose programming language).
Any thoughts. Should I just duck now?
That attitude is whats wrong with alot of computer prefessionals and give us a bad name.
Now I'm a LINUX newbie, I've got a couple of boxes running RedHat, Caldera OpenLinux and YellowDog Linux PPC, but I'm no expert. I've read the manuals and the on-line documentation, but there is alot missing.
When I have a Windoze question, I can ask around and people are helpful. When I have a Macintosh question, I either know the answer or someone on support.apple.com will help me. But when I have a Linux question...all one gets is Read The Fletchin Manual. You know what alot of those manuals are wildly out of date. I bought a Caldera book just five monthes ago...guess what...the only Caldera book then covered 1.3...I'm running 2.3 it helps me not. So what good is RTFM when TFM is out of date?
Just the other night my ISP switched our IP numbers around, a couple of hours early. My RedHat 5.2 box hung and hung hard when I tried to fire up X...then on reboot SMB took 45 minutes to start because of the IP number problem. I didn't see that in any manual. I got lucky that someone at my ISP could walk me through using PICO to edit the config files so I could get my box back up.
If I'd have asked you, I would have just gotten burned.