Cohesive Documentation Standard
I am in the process of trying to bring together the disparate (Linux)
(Although the project is not meant to be linux specific we have to start
I enclose a copy of a recent message I sent to a number of documentation
I believe it would be in everyone's best interests to work together on a
documetnation infra-structure. I have also proposed such a structure and
it up for debate.
I would greatly appreciate Debian's support in this project.
I intend getting support from several commercial distributors as this
is for the benefit of all and we need to be a little altruistic at this
The result is to be a common documentation infrastructure which would
affect any organisations ability to add documents. It would however
rules about how the documents are structured and interlinked.
Datafusion Systems Pty Ltd
Thanks to all those who responsed! I've taken your ideas on board and
included them below. Keep it up.
To those of you who haven't had a chance yet, please respond and
come on board. Now is the time before recommendations get set in stone!
What we are trying to achieve
To reiterate the point of this post:
Stop evolving documentation systems and bring all the documentation
together in one easy to access form.
Let me summarise so far:
A number ofgroups are working on mostly different aspects
of either documentation itself or documentation retrieval etc.
The organisations that I am now aware of include:
Linux Documaentation Project (LDP) http://www.linuxdoc.org
Linux Knowledge Base Project http://www.linuxkb.org
Gnome Documentation Project (GDP) http://www.gnome.org/gdp
Open Source Research Team
Open Software Writers Group http://www.oswg.org
Interest has also been shown by an electronic publisher to produce
hardcopy versions of the resulting high-quality documentation.
If you know of any (non-commerical) doc group that I've left out please
let me/them know.
I'm not even going to attempt all this on my own, if the community can't
far-sighted enough and altrusitic enough to help then the task will be
I'd at least hope for enlightened self-interest! :-)
Ultimately we could create a new mailing list covering all member
with the other existing groups as subgroups of this.
I'm currently forcing it on you for now as we keep getting new special
interest document groups and no-one appears to have shown interest in
I have joined the LDP, GDP, LKBP OSWG discussion groups and will cc this
all groups above (unless I get too many childish flames). I believe this
is one of the few
time when, what is effectively cross posting, is important to this
To those of you in advance who are going to object to my cross posting
your personal world, I'd say look around you. The best things in life,
Linux etc were made through cooperation. Maybe we all have to give a
but the result will be far better than any one group could have achieved
on its own.
I propose creating an umbrella group of which all (hopefully) the
above groups would be members.
This group would coordinate the infra-structure of all documentation,
(ie the "ideal" format, the permitted formats, the indexing/xref'ing
papaer output etc).
Once this has been dertermined (with all your input and expertise)
can, more-or-less get on with waht they were doing before, but in this
WHY am I trying to do this ?
This is a serious attempt at getting badly needed coordination. It is
NOT about control.
I am in no way interested in controlling anyone. I am interested in
a unified effort with everyones help.
I respect that each one of the above groups has slighty different goals.
This is actually good as the task is large. If we change "goals" to
"areas of responsibility" I think we could really get somewhere -
I respect the amount of effort many of you have put in,
in your own areas and groups. I do not wish to detract from that,
nor do I wish to take away anything from the individual groups.
It will take me a little while to digest all the different group areas,
with me on this, and help me out by becoming part of this discussion.
NOTE: All the following is open to comment/suggestion. This has to be a
community effort. I want to encourange informed debate, but not create
a free-for-all anything-goes environment. This is counter-productive.
Scope of Group
I've tentatively named the umbrella group "Open Documentation
Ode is also a relevant word.
(Maybe for the trekkers amongst you it should be called "ODO" Open Doc
after all ODO is very flexible.... :-) )
I wanted to avoid using one of the existing group's names
as experience indicates this leads to ownership issues.
Humans can be difficult at times. Sigh!
My initial area of interest was Linux and its software. However I
software like Gnome etc are broader than just linux. This is fine as the
document infrastructure I'm proposing supports that. It does need to be
though that the system _may_ end up being used in other areas, we should
this and not be biased against it.
In the first instance I suspect that most of the work will be done in
the linux area though.
General Design Goals
* Implementation of a Bookshelf concept
A bookshelp is a simple visual pardigm.
There would be bookselves in a range of categories (eg End User,
Only the relevant Bookshelves need to be displayed (by default)
would also be confined to selected bookshelves.
Books on any topic could be added to a bookshelf. General book
would keep books releveant to their category.
Bookshelves could be added to via a web site.
Thus a user has a list of available books.
Books would be installed wither locally, on a LAN or on the net.
All would appear to the user. (Maybe different colours etc ro
local/LAN/web access etc).
All this would be transparent to the user. Available book
pages/indexes would automatically
be downloaded and available for perusal. If a WEB based book
interesting then either the whole
book or just a few pages could be downloaded asrequired.
Indexing/cross referencing should also provide the ability to
(maybe in a pastel hue) of
information on say software that is available but not installed.
So many time users have not been aware of installed software let
software avaialble on the net.
For example if I wanted an image editor.
I could ask for a search on "Image Editors (Raster)" and be told
had say one on my system and
3 avaialble for download from the net. "Hmmm the installed one
do what I wnat. But this other one
interests me I'll download its overview page and maybe help
Looks good, ok go get it."
I have dozens (probably hundreds of programs installed on my
that I don't use.
Why are they there?
a) I might need them
b) I don't know they're there.
c) I don't need them but they were installed anyway
because I didn't
know any better
It would be so much better to install just a few "common" apps
good documetnation system.
Then have the others books/apps installed on demand.
With the bookshelf system we have access to the docs of programs
installed so we don't ahve to worry about
missing out (and often installing everything just to be sure)!
Then we'd install just what we needed.
NOTE: In the future this could also support download of
books/sw for a fee, but we
won't deal with that for now.
Books might actually consist of docuemnts that appear in several
Eg a technical chapter on PPP configuration might appear as a
document/book under Installation
and also under Administration and even uder Operating System
(internals). ALthough one has
to be careful it can be taken too far. It is not a substitude
SGI have/had a reasonable bookshelp concept, it would be useful
examine this for ideas.
* All books/documents appear to be in ONE format (from reader's
Ideally this means look at feel similar.
I'm not suggesting that we enforce a specific look and feel,
but that we conform to a set of directives (eg all titles have a
such that all titles can be made to look similar.
Recommend a default look and feel (X windowing system made the
not recommending a look and feel. The result was chaos for many
Books in a range of other formats are converted (one-time or
the common format
There are no doubt going to be some religous wars on formats.
Some would rather die than give up their favourite format.
Hopefully the converters will address some of these issues and
personal preferences to prevail.
HOWEVER. What I want us to achieve is something better for
If you'd rather defend your personal rights to the death and
part of something bigger then I wish you good luck. But such
to result in individuals and groups pushed towards the fringes
in the main stream success.
There are still people out there who think CPM (an early OS) is
best thing ever.
Well I can appreciate their right to that belief, but they
interested moving forward.
We will probably all have to give some ground for a common
If you have a technically valid issue, join in the discussion
work it out!
Working Group Tasks
The group should examine the skills and technologies of a) group
members, b) other technology
and come up with recommendations for:
Initial Documentation Areas
- split into broad subject areas right down to key
- determine the best OUTPUT technology to support
a) Good presentation
b) hierarchical doc structure
c) Extensive cross-referencing
d) High Quality Printing
- determine how best to massage existing formats into
the new format.
Identify existing documetn formats.
(Important docs would need editing, others could
processed at least initially)
Documentation "Browser" Technology.
- determine specs for a world-class doc
that fits the chosen OUTPUT technology
- take into account existing software but not be
hamstrung by it.
- might support basic browsing with existing software
advanced functions with an open-source
- include any xref/search engine
I hate layers, I hate legal documents. They often end up
Everything seems to require some legal statement though..
The author maintains ownership of his/her document and authors
must always be included.
Perhaps provide a history log of editors. If a doc becomes
must be permitted regardless. (Seems a bit hard, but I'm tired
that don't get changed). Note sure about aesthetic editing.
I'm not sure whether docs should be allowedto have individual
Maybe just permit
one or two common licenses (GNU ?). Licenses should permit free
use/printing of docs for personal
use. Non-profit distribution must also be royalty free.
distribution (eg proper publishing)
should perhaps be based on typical industry royalties and then
percentages given either to key authors
and/or the remainder put in something like an ODE non-rpofit
pay for even better doc.
I'm aware of two other doc licenses: Open Content License and
I haven't read them yet.
I'd welcome any better suggestions.
The following are a zeroth order draft of the above areas, as somewhere
NOTE: To those of you not into Linux, please bear with me I have
to start the discussion somewhere. I'd welcome contributions!
Initial Documentation Areas
Note keeps apps and utlitities separate from OS implementations
eg the unix "date" command whilst compiled for and available
Linux is also
available on any system compiling GNU tools.
It should thus be available in a cross ref under
if the linux
specific docs are removed then it ought to remain available
Note I haven't fully developed a plan for this yet.
General User Guide
Tools and Applications
Application Software Developer
Operating System Developer
There are a number of ways to divide bookshelves.
I have initially divided them based on gut feeling from years of
and problem solving for users.
Tools and Apps are separate as all levels of users need to look
Most of the bookshelves would have a combination of books that
the categories of
I guess this might be a separate bookshelf although parts of it
under Admin and
i) General Linux Installation
ii) <Vendor Specific> Installation
iii) <Vendor + Version Specific> Installation/Notes
General User Guide
a) Desktop Environment
Current (ie what is being run now)
Others )other availabe environments)
This should contain all the "getting to know you" stuff
In particular it should distinguish between the current
It is important to distinguish the current environement setup
It is too mcuh to expect a novice user to know that he/she is
Enlightenment WM on Gnome on X on Unix. Even I forget which WM
the time! The "Other" section would list alternives that the
to explore along with references to how to
Tools and Applications
a) Major Applications
Listed By Name
Xref By Category (Word Process, Text Edit, Spreadsheet,
Listed by Name
Listed by Category (Disk, Net, File Manip, Text Manip,
(Listed By Problem/Solution)
c) Man pages
Tasks by Topic
c) Internet (SLIP/PPP etc)
d) User Accounts
e) Routine Maintenace
f) Hardware Problems
z) (Man Pages)
Application Software Developer
Board Level Docs (disks, graphics cards etc)
Existing doc formats
- some HOWTOs (semi structured)
- READMEs/Changes/INSTALL instructions etc (semi structured)
- FAQs (semi structured)
- misc notes
- all command line unix commands
custom application help
- some HOWTOs
- some FAQs
- Introductory Docs
- Distribution docs (eg RedHat)
DocBook DTD (XML/SGML)
Proposed doc "OUTPUT" format
It has bveen suggested that there really isn't any excuse for
plain text any more as a plain doc can be made at least HTML
compliant with a couple fo minutes extra work.
There are also man/html converters around so the man page
format may not be as sacrosanct as once thoguht.
(being able to see good ascii output when typing "man fred"
is still a requirement though.
It thus seems feasible to have a single "meta format".
Ideally all documents would be in this format, and in time
they might be, but for now there need to be converters from
existing formats (may such beasts already exist).
Whatever standard we agree upon as the meta/Output format
it should be extensible and flexible enough to adapt to things
that we havent thought of yet.
Converters from existing formats to "OUTPUT" format
Some existing document systems are extensive enough that
conversion to a new format by hand could be prohibitive.
Some docs may be converted one-time into the new format
other (which are still being edited by other groups who
can't/won't change formats might have to be converted
on-the-fly from their native formats.
Other docs should just be changed, either by authors or group
members. Once there is enough momentum I envisage that
many existing doc. maintainers to put in a little extra work for
the good of all will become easier.
My first guess for an output format would be XML possibly using
DocBook DTD (http://docbook.org/) maybe with some extra tags.
Ease of adding/copying/indexing books
Support transfer of just sections of a book (eg specific
required to be downloaded from the web - ie don't require whole
It owuld be good to provide links from within apps to a common
Sometimes apps have internal help systems which are
a) uselss as they tell you nothing much
b) very helpful but not accessible outside the app.
Lets pursue a policy of "write once, read many" (WORM - yes I
comes from MO technology)
ie maximise the access to any documentation. This could include
activated within an application that open up to a specific part
Open to debate, but I added RPM to the "doc" list as it is
increasingly important tool for the distribution of software.
For those of you not familiar with it, RPM is a software
that pacakges up the files in a software/document/etc sub-system
an info summary and installation scripts.
RPM could be important in two areas:
1) it could be a convenient ODE Book
2) it could provide a useful method of creating a "virtual" book
Of course installed software should have documentation pages
new system but in
the meantime RPM has at least a few lines of comments about each
package and a broad category
for that package (eg Applications/Internet).
Leverage existing efforts.
Produce access to a comprehensive indexing system and permuted
Here the expertise of the OSoRT group will greatly help!
concept keywords/category (eg Text Editors->Graphical" or
The Gnome Help Browser might be a good GUI software vehicle with
to start experimenting.
Initial GUI Thoughts
I'd like to see a doc browser a bit like one of MSoft's (eeek!)
development doc system.
The key idea is a panel at the left with a tree view and the
panel at the right.
Might also have a separate document/book-chapter/section broswer
quickly pick out sections
in the current book.