Alternate wiki-implementations

From Consumerium development wiki R&D Wiki
(Redirected from Wiki code)
Jump to: navigation, search

Wiki code enables the Research Wiki and Publish Wiki. Choices we make about it will guide the Consumerium Software License and likely also the software development process - probably in ways we can't easily change or undo later. With so many alternate wiki implementations out there, it is important to sort through the advantages and disadvantages of each.

This starts by considering the wikitext standard that our present text is in: Currently 3 out of 20 of our registered users are registered MediaWiki developers, which makes our percentage of developers among users 15%, which is likely the highest figure any public MediaWiki installation can boost so that is an good incentive to try to adapt MediaWiki for our use over other wikis. However they might just be here because we are using MediaWiki, so, it is important to make clear that one of the things the R&D Wiki is doing is choosing what technology best fits our hardware requirements later.

Also, a good software development process requires shifting to a new development base when moving from R&D to production code - most studies show that it is very very bad to use the same platform and the same libraries, for instance, as one used in the R&D phase, in real production.

So, there is ample reason to believe that we might move off MediaWiki later, or reserve it only for R&D use. There are three leading candidates, and a few dark horses listed afterwards. It seems likely that we'd ask those who want to be Consumerium developers to work on wikitext standards and on soliciting and forwarding end user feedback better, starting with our own MediaWiki modifications requests. Consumerium users should not have to do anything but list these here. Similar pages for TikiWiki modifications and MoinMoin extensions can be created, and which one meets our needs can be more of a competition. In most cases, the features that must be added are different for each package, since they start with different feature sets.

Also APC Action Apps might become important to integrate, since they have broad use among nonprofits. One might for instance be written to support some common barcode readers, in cooperation with Adbusters or another of the essential projects which recognizes barcodes as being as critical as we think they are. Maximum cooperation and out-sourcing should be encouraged.


MediaWikiGetWiki
  • Not humanly possible for end user feedback to reach developers reliably, though Meta-Wikipedia has some channels experienced trolls can exploit
  • PHP based
  • Proven to perform well under heavy load - but with hard limits
  • Only rational basis for wikitext standard given over one million articles in fifty languages
  • Working now for R&D purposes
  • Dedicated developers heavily involved in content problems, although they are in general Wikipedia content problems
  • Readable documentation, thanks to the above
  • supports MySQL only
  • already converted into the TomeRaider format for off-line browsing on many handheld devices

GetWiki is a fork of MediaWiki under CC-nc-sa

  • same de facto wikitext standard
  • Auto-import of Wikipedia articles via XML using PHP's Expat Library
  • Well-formed XHTML, replacing deprecated HTML 4.0 versions
  • Improved use of CSS styles (colours, positioning)
  • Cleaner output to the browser, PHP code underneath
  • More integrated and effective use of CSS2, beyond font colours
  • More intuitive User Preferences page
  • Clearer labeling and checkbox placements
  • Fixed user login retention and other fixes
  • Better use of licensing notices on pages
  • Image thumbnail display
TikiWikiMoinMoin

Dark horses include VeryQuickWiki (a Java wiki very easy to set up and run without problems), UseMod (only advantage is that it dumps XML output, very very very important until there is a real wikitext standard), and Zope (also dumps XML, and is in use at other essential projects like ACT.Greenpeace.org. There is finally Drupal which has many advantages and includes XML output as well.

Also Microsoft wiki will likely be out eventually, and some peer2peer options relying perhaps on XForms later. Microsoft Internet Explorer XML Support is an important constraint on which of these features can be used at all. It may be a wise tradeoff to support only MSIE for certain user roles, if these are always taken on by well-equipped well-supported people who can be far more easily supported by relying on IE's XML, than any other method. By the same argument, if Opera or Netscape does something uniquely well and only two or three people need that capability, requiring those might also make sense for some roles. For ordinary end users the APC Action Apps baseline browser capabilities should probably be adopted as Consumerium's own.


See also: