Evaluate new static search engine using AJAX method

[nexero]

In order to push forward into hosting the catalog files on github pages #6, we need to do the search differently.

I can think of three possible approaches:

• Convert all pages into HTML and host them on Github Pages:
  • Benefits:
    • much faster page rendering due to rendering while still loading page, not possible with the XML/XSLT approach
    • ready to use fuzzy search throughlibs like lunr.js
    • better full text search
    • no CORS restrictions while previewung the file on a local machine
    • as we are not limited by browsers XSLT implementaions, we can use XSLT >1.0
  • Caveats:
    • search is too generic regarding terms that occur on every page
    • tag search must be implemented differently
• Use old XML/XSLT approach:
  • Benefits:
    • better preservation of semantic data structure
    • fast search through files by using (existing) XPath expressions
    • very fast and robust search through tag names
    • smaller files
  • Caveats:
    • bad full text search
    • fuzzy search is hard to implement in XSLT 1.0
    • CORS restrictions, so no preview on local machine possible without a running web server
• Kind of hybrid approach (with most of the corresponding benefits mentioned above):
  • Host XML files beside the HTML files
    • Caveats:
      • more files in the repo
  • Embed XML in HTML files
    • Caveats:
      • much more data to load when searching through all files
      • might be slow on low-end devices

I'd like to use the hybrid approach and host all XML files beside HTML files, but implement fuzzy full-text search using XPath 2.0 and/or XQuery, because we can control the search result better then.

@scymtym, @warp1337: What do you think?

[warp1337]

Phew. Need to think about this.