Re: [Refdb-users] phpweb interface HAS a problem for me, and for you ?

[Markus H. <mar...@mh...>]

[...second attempt...]

[cc'ing the list as this might be of common interest...]

Hi Dan,

Daniel O'Donnell writes:
 > I was thinking of setting up a "Preferences" section that would allow
 > users to place some cookies saying how they want certain things to
 > behave. It could be added there. Or we could build an initial "install"
 > page (like mediawiki does) that allows administrators to specify what
 > dbase they are using and other things like that.
 > 

Another option is to use the refdb-init script which could perform
these customizations once it is at it.

 > Actually I was thinking along the same lines and have been debating the
 > best way of doing it. What I was thinking was a) using drop down lists
 > (collected by running getau, getkw, etc) whenever users are supposed to
 > enter information that might already be in the dbase--with a blank text
 > input field beside the drop down list for info that is new to the
 > database. This would be in the addref page and involve AU, JF, JO, KW,
 > and maybe BT.
 > 

Emacs refdb-mode handles this in a similar fashion. You can enter
authors, keywords, and periodical names with tab completion, or just
type in something new.

 > and b) building links so that clicking say on an author name actually
 > runs :AX:<author name>.
 > 

refdb-mode is a little more advanced here, but this is possible only
because it is part of an editor. You can highlight any word or phrase
and use it as the target of a query (author, citation key, title,
keyword, whatever).

 > The second isn't hard, I don't think, though my PHP is still new. The
 > first wouldn't be either, I don't think, but generating them on the fly
 > might cause speed issues. I'm thinking either adding an "update indices"
 > at the beginning and end of the session with perhaps automatic adding
 > whenever something is added or deleted. The drop down lists would then
 > be premade.
 > 

refdb-mode updates teh internal completion lists whenever a command is
run that alters the database (add|update|delete). I don't know whether
these lists can be made persistent in a similar fashion in
PHP. Updating these lists is a speed issue with larger databases and
slow database engines.

 > The only thing that would affect the design of this, I think, is whether
 > or not we decide to allow multiple personal lists (which I'd like, it is
 > a feature of both Procite and Refworks).

I've been thinking about this for a while. I don't think it makes
sense to duplicate a feature that is already possible with extended
notes. Instead, I figured it should be possible to implement personal lists
transparently by abusing extended notes under the hood. Consider the
following command:

pickref 13

This would run the equivalent of the following client commands on the
server side:

addnote <empty template with title=username> (unless that note already
exists)

addlink :NID:=<id where title=username> :ID:=13

dumpref 13 would run the equivalent deletelink command. Up to here,
everything is entirely transparent to the end-user and perfectly
backwards-compatible.

To add a reference to a different personal list we'll need an
extension of the pickref command:

pickref -P mylist 15

This would run the equivalent of these commands on the server:

addnote <empty template with title=mylist> (unless that note already
exists)

addlink :NID:=<id where title=mylist> :ID:=15

So you'd be using extended notes without even having to know how to
write them. However, if you want to, you can of course retrieve these
notes with the getnote command, add a description or a few keywords,
and treat them just like any other extended note.

As easy as it sounds, please don't expect this feature to pop up
tomorrow. There's more to it on the programmer's end, like user
permissions and orphan cleanup.

 > I'm also still not sure how to
 > search (commandline or php) for words with apostrophes in them.

A quick test indicates that there must be some error in the code that
is supposed to escape and unescape apostrophes. I'll investigate this.

 
 > And
 > finally, I'm not sure that call number on AV is working out--sometimes
 > things have both a callnumber and AV. I wonder if we should decide on a
 > user-defined tag for it?
 > 

As always, we could make this a configurable option.

regards,
Markus

-- 
Markus Hoenicka
mar...@ca...
(Spam-protected email: replace the quadrupeds with "mhoenicka")
http://www.mhoenicka.de