[Indic-computing-devel] X Test Suite failures in NCST IndiX

[<jk...@Fr...>]

As expected, running the font handling portions of the X test suite
against NCST Indix results in test failures in font drawing.

Here is a summary:

                             IndiX                      XFree86
                             ~~~~~                      ~~~~~~~

[OS]                         FreeBSD 4.4-STABLE         FreeBSD 4.4-STABLE

[source base]                X403-indix-src-*          (X336 public release)
                             (rohini.ncst.ernet.in)    (xfree86.org)

[server version]             4003                       3360

[note: the vendor string for IndiX still records the server vendor as
 "The XFree86 Project, Inc." (as reported by `xdpyinfo') ]

-- IndiX bugs --
XDrawString                  39P,1F                     40P,0F
XDrawString16                40P,0F                     40P,0F
XDrawText                    42P,2F                     44P,0F
XDrawText16                  29P,14F                    43P,0F
XQueryTextExtents16          1P,2F                      3P,0F
XQueryTextExtents            1P,2F                      3P,0F
XSetFontPath                 2P,1F                      3P,0F

-- common bugs in both code bases --
XQueryFont                   1P,2F                      1P,2F
XLoadQueryFont               3P,1F                      3P,1F
XListFontsWithInfo           4P,2F                      4P,2F

[Legend: aP, bF means: 'a' passes and 'b' failures.]

Additional Notes:

 - The test run was conducted using the 'pure framebuffer' X server
   "Xvfb", to isolate the test from any bugs in graphics card specific
   driver code.  

 - Only the test cases that deal with glyph rendering were
   invoked.

I remember raising the issue about the protocol (non)compliance of
NCST IndiX a year or so back.  I am disappointed to see that the NCST
IndiX web page has STILL no mention of the its non-compliance to the X
Window System standard, leaving its users to figure this out the hard
way :(.

In my commercial X servers writing days (a decade back), even ONE X
test failure was enough to stop code shipment.

Koshy
<jk...@fr...>