[Pmwiki-users] Re: Re: Re: Organization of pages (was: Hierarchical groups)

[Christian Ridderström]

Oops.. thought I'd already sent this. Since I can't remember why it was 
still in my "postsponed" box, I'll just send it.

/Christian

On Sun, 13 Jun 2004, Patrick R. Michaud wrote:

> On Sun, Jun 13, 2004 at 11:03:36AM +0200, Christian Ridderstr?m wrote:
> > 
> > > > It's probably a good idea to think of the pages as a database (regardless 
> > > > of how it's implemented), and look at how to structure it from that point 
> > > > of view.
> > > 
> > > PmWiki 2.0 does just that.  
> > 
> > Ok. Do you have a reason for not giving pages a unique ID? From my
> > experience with databases I'd say that's almost a requisite?
> 
> Well, pages *do* have a unique ID, the page name is a unique ID.

Yes (when you include the group name in "page name"). But this ID is not
constant if we allow pages to be renamed (more than just changing the
title).
  
> The question (I think) you are really asking is why not give pages
> a unique ID other than the page name,

No, I mean in *addition* to their page name. It should be as simple as 
adding a field: "ID=2323" where the number is increased each time a new 
page is created.

Btw, I'm more and more starting to think of the page name as the "URI" of
the page. And further more, I can think of cases where I'd want the same
page to be accessed through the same URI. (We've already discussed using
[[include:]] and [[redirect:]] as a solution for this)

> and the answer is that I then have to come up with a set of tools and
> interfaces for a wikiadministrator to be able to manipulate the pages in
> the database (no matter how it's stored)--i.e., removing groups, copying
> pages from one wiki to another, and other things that can be done within
> the filesystem now become much more difficult and require a lot of extra
> scripting and support.

I don't think we are talking about that much work here (and if necessary I 
could write it for you --- that's how much I think ID's are needed).

> I don't see the benefit (version control of page names) as being
> anywhere close to worth the cost in terms of ease of maintenance,
> importing pages, exporting pages, backups, etc.

I just realized that with an extension, version control of page names
would let you deal with renamed pages (that don't have a redirect).
Would that be useful?

> Using IDs other than pagenames throws out a lot of tools (ftp, rm, mv,
> del, cp, etc.) that are now available for administrative purposes, or at
> the least makes them more difficult to use.

* 'rm' will still work (the pages i definitely gone though)
* 'mv' will still work (the page has been renamed though)
* 'cp' would need extra work aftwards (*)
* Restoring a page that has been backed up will still work
* Importing a page from another wiki would need extra work afterwards (*)

I don't think it's that difficult, and I'd be fine with having to choose 
between manual manipulation of the database and being guaranteed unique ID 
numbers of the pages.

/Christian

(*) In this situation a page may have been added that has the same ID as
some other page. However, I think this could be handled automatically by a
"cleanup" script, that wouldn't even have to know which file was
incorrectly added. It could do something like this:
* Scan all pages for their IDs and generate a list
* Check the list of IDs for duplicates
* Give the more recent file a new ID (possibly logging the change in 
  both files)





-- 
Christian Ridderstr?m                           http://www.md.kth.se/~chr