> So far I've understood that .device is a native roff macro It's not a macro, it's a (built-in) request.
> that passes its argument through to the output roff wrapped in > something that designates the postprocessor that this is a "raw > passed through argument" No. As the name `.device' suggests, it is mainly intended for communication between troff and a postprocessor. > hen the postprocessor picks this up (for instance grops picks it up, > but only accepts it with a ps: prefix) and acts accordingly. Yes. > So the change would actually be needed in grops to account for a > sequence such as "ps:papersize" and write out the according > ps-instructions that you more or less mentioned). Yes. > Is there perhaps a generic (for all sorts of devices) groff > command/sequence that passes the data through as a binary blob, > meaning that it goes untouched by both, roff AND the postprocessor? > So one could eventually specify the PS-sequence in the groff > sourcecode (like a stronger \X)? Yes and no. Look up this thread: http://lists.gnu.org/archive/html/groff/2008-02/msg00019.html http://lists.gnu.org/archive/html/groff/2008-03/msg00000.html Since grops limits all inserted stuff to be encapsulated page-wise, it is not possible to insert high-level DSC stuff. In other words, it's really necessary to introduce a new \X'ps: ...' sequence for manipulating the paper size. Werner
