On Tue, 16 Dec 2003, John Sechrest wrote: (01)
> 1) does every page have a "did I answer your question" question? (02)
Yes, but the data is gathered to a log that is not processed
(it's on the list of the low priority things to do, which is
crazy...) (03)
> 2) Are the other links generated automatically or by hand? (04)
The links are generated from document id at the time of
transformation in the presentation layer, but those document id
references are put in by hand. (05)
> 3) What programs can reason over the knowledge base? (06)
None that I'm aware of, if by reason you mean something akin to
the semantic web dream. (07)
> 4) How do you export the knowledge base? (08)
It is not currently exportable as there is no way to serialize a
Document object (a Document is a text and all its metadata) in
any reasonable format (you could Storable the perl object, or
xmlify the hash which represents the object, but what have you
got then, not much). (09)
> 5) What happens when you want to include images? (010)
I'm not quite sure how this works, but the KB provides support
for including image, audio and video objects with Documents. This
feature has not proven sufficiently valuable in the context of
the current system for it be used. (011)
> 6) What happens when you edit a page? how does the approval process work? (012)
A designated kb editor has special permissions on the editing
system (a Solaris box with Oracle and a bunch of code). They open
up emacs and start up the editing mode and either choose to
create a new text or a new version of an exiting text. That text
is RCS version controlled. All texts have an owner that is
designated during the creation time. The owner is a role account.
A person operating as that role, or a specially designated
editor, has the power to give a document a stamp of approval
(soa its called in the local jargon, which confused me from my
dns admin days) and "make it live". (013)
> 7) Can you assign permissions/access patterns to subpages?
> is there a delegation process? (014)
There is no such thing as a subpage. Everything is a single page
(this is a central issue in my "you can't call this a content
management system" problem). Access to pages are controlled by
domains. Domains exist in a polyhierarchy and are usually
associated with class C or class B subnets but can be controlled
by single hosts or (coming soon) user based authentication can
grant domains of access. (015)
A document can have <kbsecure> sections or who attributes on many
elements which control display of that thing to a particular
domain. (016)
> 8) Can you simulate everything you need for KB with a wiki?
> would it be harder or easier than the current KB? (017)
No, but why would you want to? They serve different purposes,
have different intentions. (018)
The primary issue is probably content ownership and lack of
sufficient metadata for management purposes. (019)
> 9) Can you simulate everything you need for a KB with a collection
> of info files? What would be missing? (020)
I've not explored this option well enough to comment, although I
feel info, for reason I can't understand, to be horrible. (021)
I think this is because I've always seen it as "The GNU project
maintains documentation in the info format. This man page is no
longer being maintained." Or something like that. To which my
reaction is "eat me, give me my damn man pages." (022)
(This is not such a problem these days, but a few years ago it
was.) (023)
> I am guessing that there is a lot of hand work and little automation.
> Am I guessing right? (024)
Content generation is hand work, but that's good. That's what
makes the content have some semblance of quality and color. When
the kb travels to conferences the thing that is stated over and
over again by the audience is their surprise that IU actually
managed to pull of the content gathering and generation. That's
the hard part. The software, while complicated, is light in
comparison. (025)
Now that the kb is so embedded in iu's it infrastructure,
changing it to meet modern external standards is rough. (026)
>From my point of view I don't see much overlap between the kb (as
IU's done it), wikis and info pages. They serve much different
purposes and audiences. The historical context for the kb is for
it to be a chromified faq repository. I'll let someone else jump
in on wikis and info? (027)
--
This message is archived at: (028)
http://collab.blueoxen.net/forums/cgi-bin/mesg.cgi?a=tools-yak&i=Pine.OSX.4.53.0312161605330.2229@nitrous.local (029)
|