Implementation report on generic URI syntax

The test page used was <http: www.ics.uci.edu="" ~fielding="" url="" test1.html="">
It has been tested on:

  • Mozilla/4.03 [en] (X11; U; SunOS 5.5 sun4u; Nav)
  • Lynx/2.7.1 libwww-FM/2.14
  • MSIE 3.01; Windows 95
  • NCSA_Mosaic/2.6 (X11;SunOS 4.1.2 sun4m) libwww/2.12

The following features were tested:

  • Hierarchical URLs
    • Implemented in http, file, and ftp URLs on all tested applications
  • Nonhierarchical URLs
    • Implemented in mailto and news URLs on all tested applications
  • Escaped encoding
    • Implemented on all tested applications above
  • Fragments
    • Implemented on all tested applications above
  • Userinfo
    • Implemented on Lynx/2.7.1 and Mozilla/4.03 (did not test others)
  • Host/port
    • Implemented on all tested applications above
  • Path
    • Implemented on all tested applications above
  • Relative path resolution
    • Lynx/2.7.1, MSIE 3.01, NCSA_Mosaic/2.6 all parse normal examples
      within compliance limits. There are still a couple cases where
      Mozilla/4.03 doesn't handle unknown schemes correctly, but people
      at Netscape claim they will fix those as known bugs once the RFC
      is done.
  • Semicolon within a segment
    • Used in ftp and imap URLs (that is, ";" is reserved for those)
      Also used in file URLs on VMS systems, but I don't count that.
  • Query
    • Used in http and wais URLs
  • Various types of Base URL
    • All tested applications support file and http base URLs.
      All except Mozilla/4.03 supported a test of the "fred" URL
      scheme as a test of an unknown scheme.