[Epsonepl-users] broken IJS device in ghostscript 8.0, and the fix (ghostscript-Bugs-677383)

[Hin-Tak L. <hin...@ya...>]

I have found a fix to the ghostscript 8.0 bug that I
reported 2 months ago
(http://bugs.ghostscript.com/show_bug.cgi?id=677383).
I believe it probably affects all IJS plugins to ghostscript
if one wants to use ghostscript 8.0, hence I am CC'ing
the other IJS-based projects. Since gs 8.0 is not yet
widely used, I suppose we just have to be aware of it
and hope the next release of GNU ghostscript comes
with the fix. This bug affects ghostscript 7.30-7.32 onwards
up to and including 8.0 and was introduced by the
colour-related DeviceN restructuring around 7.30.

The fix is as follows (require recompiling ghostscript
7.30 to current) - about 15 lines counting backwards
from the very end of "ghostscript/src/gdevijs.c":
=====================
                        maxvalue = (1 << bpc) - 1;
                        dci.num_components = components;
                        dci.depth = bpc * components;
                        dci.max_gray = maxvalue;
=====================
A line "dci.max_components = components;" needs to be inserted.

(note it is 'max_' in *addition* to the 'num_' member). Without
it, "max_compoments" stays at 3 initialized elsewhere.

"max_components" was introduced in gs7.30. "num_components"
dates back a lot longer, although they seems to be related.
I cannot find any documentation on how "max_components" is
used (it only appears in about 5 places in gs 8.00, whereas
"num_components" occurs in almost every file), and it seems to
be set to "num_components" almost all the time. Hence the
fix. Admittedly, I don't completely understand why the fix
works, but it makes "epsoneplijs" works correctly. I would
appreciate if say, Raph Levien, can kindly explain what
"max_components" is supposed to do? And is this the correct
way of fixing the bug?