[pmwiki-users] Blog proposal

[Patrick R. Michaud]

On Wed, Jan 11, 2006 at 12:59:38PM -0500, Martin Fick wrote:
> > But the idea of making it easier to create trails has lots of
> > merit.  But I'm wondering if this means that the generic form of
> > trails becomes the new way to handle categories as well -- i.e.,
> > a page with category markup automatically adds itself to the
> > corresponding category trail.
> 
>   This is an intriguing idea.  Would it also create any such
> Category pages if they don't exist?  

Yes.

> I guess this
> effectively would put the indexes directy on the category
> pages.  

Yes.

>   Along the line of indexes,  from the changelog, it looks
> like you removed the trail prev and next pages from pages on
> a trail. This is something that I requested, but I am
> curious now as to how it worked and what specifically made
> you decide to remove them?

Well, it had always bugged me that pages were saved with
links to their prev/next pages on trails, since the page
itself didn't actually contain a link to them (and since
editing the trail page could cause the links to be invalid).
However, until the 2.1.beta series there wasn't an easy
way to keep those links from appearing in targets= because
of the way links are generated.

However, there's now a $SaveAttrPatterns array, which
is a set of transformations to be made prior to calling
MarkupToHTML() as part of computing the page attributes.
Trails.php simply uses this array to remove any
trail markups from the page before the targets=
attribute is computed.  Similarly, pagelist.php uses
the array to remove calls to (:pagelist:) and (:searchresults:)
so that they aren't called when evaluating the page.

>   My big aversion to indexes though, still has to do with
> giving up the external management simplicity of flat files.
> One way that I might feel less impeeded by any such schemes
> would be if there were a way to call pmwiki externally from
> a shell with a pagename and have it update everything that
> is needed for that page.  

Why would it have to be done from the shell...why not from
the web?

    http://www.example.com/wiki?n=Group.A&action=reindex

(Presumably one could achieve the same result from the shell
by using a command-line browser such as lynx or wget.  Heck,
even "firefox" would work for this.  :-)

Pm