[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re^8: Debian Metadata Proposal -- draft rev.1.4

Am 09.07.98 schrieb Marcus.Brinkmann # ruhr-uni-bochum.de ...

Moin Marcus!

MB> Marco, I get the impression here that you are objecting for the sake of
MB> objection.

Please remember your mail: "I don#t like dot files!".

MB> (a) I don't care what dhelp needs and what it doesn't. I want to get the

But I do.

MB>     design right. Dhelp was a preliminary hack. At the moment, Debian

No, dhelp is no hack. It#s a working system.

MB>     doesn't have a working documentation system, and so we have the
MB> freedom to     design a good one.

We have to working systems, but I don#t have a problem to support new  

MB> important ones. What is a "real objection" for you, Marco?

I#ve posted them some days ago.

MB> > Where#s the connection? Your URL points to the file on the local
MB> > filesystem and not to the file on the system where the docreg file is
MB> > installed. Ergo no solution.
MB> >
MB> > (Are you sure that your URL is correct? netscape uses file:/usr/local/?)
MB> It is indeed valid. And, please, Marco: Make yourself informed about
MB> identifiers, URLs and URNs before you participate further in this
MB> discussion.

I will do so, but file://localhost/ is no solution.

MB> This is not at all. slow. It will start a bit slower and then will go back
MB> to 100% as soon as all packages made the transition. And this is only for

Wrong, we can#t assume that the user will use slink (for example) packages  
only. Some people need always old packages (see libc4 problem).

MB> It is indeed nonsense. Please read my summarizing mail carefully. Think
MB> about what an identifier is. Try to seperate absolute identifiers and
MB> relative identifiers. They are not necessarily filenames, Marco.

That#s right. But your proposal will use the path/filename as identifier,  
too. There#s no real difference between both solutions: both solutions are  

The old proposal was a lot of better.

MB> > Maybe it#s not a good idea, to use the filename as identifier. Our old
MB> > solution (doc-base) hasn#t got this problem.
MB> It is INDEED a BAD IDEA to use FILENAMES as identifiers.
MB> And it is EXACTLY what YOU are trying to do with your "relative to the
MB> actual directory" mis-concept.

Where#s the difference to your solution? Do you need some examples? Here  

  1.) The doc-linux-html maintainer has to move the HOWTO from
      HOWTO/ to HOWTO/html. This will change your identifier, too.
      With both solutions the links (doc-linux-{de,fr,..}) will be

  2.) I would like to add one URL twice. This is not possible with
      both solutions:

         Identifier: www.debian.org
         Lang: en

         Identifier: www.debian.org
         Lang: de

MB> The point is that they are STILL unique, when we use the path
MB> /usr/share/doc, /usr/doc,

Again, the user may have several other directories and then they#re not  
unique! This is a bad design. For example Adam suggested to search  
www.debian.org/doc, too. If the maintainer of www.debian.org installs
the same packages like the local maintainer in an other version you will  
have a problem.

Searching several directories is bad. The identifier should be unique (we  
should use a URL or filename!) and doc-base should know the right  
directory of the files.

MB> > Why should we use a file name as identifier? This is always a bad
MB> > solution.
MB> You want to use a filename as an identifier, we want to use a unique URL
MB> (maybe later URN) as identifier. Those are different, and the whole
MB> discussion would improve greatly if you would learn this difference soon.

?? Where#s the difference URL <-> filename? There is no difference in our  
discussion. And something like file://localhost/doc is not unique. Dublin  
Core itself suggests to use the homepage URL and not a local file: URL.

And the usage of the homepage is a problem, too.

At the moment there#s a real problem in the WWW, you don#t have something  
unique like the ISBN number. That#s why people are talking about systems  
like RDF.

We need a unique identifier, right. We could use something like  
<package><document>. But this should be only a identifier! We should use  
that as filename and of course Adam#s proposal use it as filename/URL.

In the WWW the DC informations are included in the HTML file! That#s why  
you don#t need a filename. And if you use it for libraries you could use  
the ISBN number as identifier and you don#t need a filename.

MB> > Really? Please tell me how I can add for example
MB> > /usr/lib/doc-base/foo.docreg to a tar archive. 99,9% of all tar archives
MB> You need an install program that moves the file, sets the URL identifier
MB> (note: NOT the filename, see above) correct accoring to the document
MB> location and you are done.

Great solution :(. Why should we do that if it#s not necessary?

MB> BTW: Our main concern should be Debian packages, not vendor packages.

I don#t think so. Think big!

cu, Marco

Uni: Budde@tu-harburg.de           Fido: 2:240/5202.15
Mailbox: mbudde@hqsys.antar.com    http://www.tu-harburg.de/~semb2204/

To UNSUBSCRIBE, email to debian-doc-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

Reply to: