> However, I think you need the ijs
> driver, which is not built by default (it probably should be).
> To enable it, edit /sys/src/cmd/gs/mkfile to add ijs to the
> device list and then rebuild using the instructions in the
> mkfile.
I thought I'd built gs(1) this way with proper ijs support,
but `{gs -h} quickly made me realise otherwise. It was
a hoax; there's quite a bit more that's needed in order
to build with ijs support: for one, the ijs/ directory is
missing, which contains the proper headers and other
sources; other edits need to be made to the mkfile,
as well as to src/plan9.mak. Using Prof. Okamaoto's
sources from a while back, I've finally got it built,
and I'll test whether I can now print something (still
using the old hpijs 1.5, as it supports OfficeJet 5500 -
I figure, "close enough").
Furthermore, changes needed to be made to
/sys/lib/lp/process/gspipeijs, which was clearly geared
for Prof. Okamoto's own setup (notice the switch
statement: $GSOPT is nil for any model numbers other
than the ones specific to his HP PSC printer). Small,
but necessary and elusive (the output of `{gs -} from
gspipeijs is very misleading). I think $GSTMPFILE also
needs to be changed.
I hope everything in /sys/lib/ghostscript is proper with
regards to this; I would really dislike problems due to
missing files there. I don't know anything about them.
My changes for building gs(1) with ijs required a lot of
out-of-the-garden stuff. I'm guessing the problem has never
before occurred because the code has been left dormant
and no one has made use. Thus, perhaps increasing the
size of gs(1) sources in distribution with the appropriate
data and files to build with ijs support is unnecessary.
Most of it is just taking what Prof. Okamoto had already
done; if anyone would like, I can document these changes
to gs(1) sources.
Best,
ak
