| WebSeitz/wikilog |
| Sister Sites |
|
| last edited by BillSeitz on May 14, 2008 10:28 am |
Context:
you are a reader/writer of multiple Wiki Space-s (see Wiki Types)
other members of these spaces are members of overlapping sets of Wiki Space-s
these various clusters of spaces "cover" overlapping topics
you would like to avoid creating copies of the same content in every space
Forces
Solution
encourage each site to support Automatic Linking (via Smashed Together Words), Page Name As Url
automatically link across spaces to pages matching a given Wiki Word whether such a page exists locally or not
Twin Pages: for the current page, show which Sister Sites have a page with the same Wiki Name
Near Links: when a page mentions a Wiki Name which doesn't exist in the local Name Space, but that Wiki Name exists in various Sister Sites, make it easy to see that list and follow one or all of those links
MeatBall:WikiEnginesWithSisterSites , MeatBall:SisterSitesImplementationGuide
MeatballWiki:MetaWiki - it would be interesting to query this to find out which wikis have the most shared names with you.
Here Ward Cunningham suggests that each WiKi maintain a [URL] which lists all its pages, 1 per line, with MIME-type text/plain. (He doesn't specify order - also, I wonder whether the code he's used to can handle having additional data on it - this might allow multiple uses for the same [URL], like Touch Graph lists, Site List Txt (see Web Robot), etc.)
Sept'03: MeatballWiki:MetaWiki can be used to provide a list of Sister Sites for every page on the wiki you submit to it. (See section "How to use [Meta Wiki] to implement Twin Pages")
[Meta Wiki] seems to build its listing by scraping MeatballWiki:AllPages listing for each member wiki. Should that be returning HTML or text?
It appears that ZWiki:SisterSites used ZWiki:MetaWiki at some point, but then stopped. Was the list, just of matching pages, really 650kb back in 2000? And it seems like the code was based on querying the pure text return from [Meta Wiki], rather than building a PyThon dictionary (or ZoPe Catalog, or something).
Should this be done in the client? Or maybe a Wiki Proxy?
Who should define the set of sites to search/link - the author, or the reader? Maybe the author suggests, and the reader decides. (Kinda like a blogroll with subscribe-links, e.g. on Jon Udell's sidebar.)
how does that differ in Intra Net environment vs public spaces? or private spaces (maybe I want my public wiki to make it easy to jump to my private wiki, say, to look up contact info on someone)
I'm working on some code to implement this for Pikie Pikie, which is in Python, so it may be of some use to ZWiki-folk once i get it working. See Pikie:Twin Pages (or is it PikiePikie:TwinPages ? pikie.darktech.com is down temporarily anyway, so for now see http://pikie.webbing.nl/cgi-bin/pikie.cgi?TwinPages ). --John Abbe
Some possible approaches for me - various scenarios
(general reminder: have to catch when a site deletes a page - or maybe not just be sloppy)
Using [Meta Wiki], just for this WikiLog
note that this approach doesn't let you link to sister sites for nodes that you mention but don't locally have
could walk catalog of forward-links to build longer list of local pagenames to submit
need way to save/manage list of which wikis I want to show (assuming I don't want to show every single [Meta Wiki] space)
just add another persistent property to each page object
doesn't work if I want to link to pages that I don't locally have
Handling this WikiLog without [Meta Wiki]
grab each site's listing url, building dictionary (which would be of every page at any of those sites, not just those matching local pages); then at end go through and turn into persistent properties
Supporting all locally hosted Wikilogs Com sites (assuming not satisfied with having replicated process/data in each individual wiki)
start by building a list as a union of all pages on all sites
submit that list to [Meta Wiki] to get matches; save to local dictionary and save it (don't even know how to do this in ZoPe!)
then have a function to query that dictionary: should be pretty fast
Running a small [Meta Wiki]-like service for a small number of friends' wikis hosted elsewhere.
Supporting a number of Teamflux Com sites as part of a single enterprise.
do they need to be in contained in a single parent folder?
other????
| User Options Recent Changes Help Page |