FULL DDX MODULE INSTRUMENTATION -- THE FIRMWARE NAMES THE FAILURE.

1) WHERE THE EXPORT ACTUALLY DIES
   The trace showed the cleanup/return block's call 0x7E75EA8 returning 2:
     after 7F67F38 a8=00001441
     after 7E75EA8 a8=00000002   <-- returns BadValue
     after 7EA45FE a8=00000000
     after 7E7FE96 a8=00000000
     RET a8=00000002
   file 0x7E75EA8 - 0x7E40000 = **file 0x35EA8**, which is the DDX render entry
   (the CALLA is made with the constant 0xa6000, the render-params table).

2) THE DDX MODULE'S OWN DIAGNOSTIC STRINGS (extracted from vt1200_v22.bin)
   file 01C6DF  "error allocating renderparams block"
   file 01C775  "RenderSetupPipe:      %d ms"
   file 01C877  "RenderSetup:          %d ms"
   file 037156  "DDX_RENDER: unsupported compression scheme"
   file 037185  "DDX_RENDER: renderptr->src_base == NULL == %x"
   file 03722E  "DDX_RENDER: error allocating tmpudp or smiconvert to dst_photo"
   These are the three DDX_RENDER failure modes, in that order.

3) WHICH ONE FIRES -- MEASURED
   Instrumented inside DDX_RENDER (file 0x35EA8):
     CLS   0x35F0E  reads  *A11(0x630)  (renderptr byte 198)
     F610  0x35F2A  reads  *A9(0x610)   (renderptr byte 194)
     ERR2  0x35EF2  -> prints string 0x3F3B8AB0 = "unsupported compression scheme"
     ERR610 0x35F30 -> prints 0x3F3B8C28 = "renderptr->src_base == NULL == %x"
   Live run (genuine DEC client, export with and without a colormap):

     DDX class=85   (0x55)  -> DDX_BADVALUE_class -> DDX_RET a8=00000002
     DDX class=115  (0x73)  -> DDX_BADVALUE_class -> DDX_RET a8=00000002
     DDX class=123  (0x7B)  -> DDX_BADVALUE_class -> DDX_RET a8=00000002

   The dispatch validates renderptr byte 198 against {0,1,2,3} (mapping to
   5/0/1/3) and rejects anything else.  The values read are **0x55 / 0x73 /
   0x7B** -- the heap fill pattern and heap residue, i.e. renderptr byte 198 is
   NEVER INITIALISED.  So the reported failure is literally
   "DDX_RENDER: unsupported compression scheme" with a garbage scheme.

4) WHAT THAT MEANS
   The render-params block (allocated by the export path; the firmware even has
   a message for failing that) is handed to DDX_RENDER with its compression-scheme
   field unwritten.  Everything the client can set -- transport mode/scheme,
   photomap compression (XieK_PCM=0), levels, organisation, drawable, GC, depth,
   geometry -- has been verified correct, and the imported data is present
   (element+0x28 = 0xFF white / 0x00 black).

   Passing colormap_id (DefaultColormap) instead of 0 changes the garbage value
   but not the outcome: still uninitialised, still rejected.

5) STATE
   No client-chosen image displayed.  Window untouched; white vs black captures
   differ only by Terminal Manager figures and cursor.
   xshow (core X) is the only viewer that has shown an image.

This is the firmware's own diagnosis, not an inference from absence.
