Re: [Gnome-print] Encoding in postscript output

[Chema Celorio <chema ximian com>]

Scott Gifford wrote:
> My plan is to fix up the PDF encoders so they can work with blocks at
> a time, instead of requiring the whole string to be read in, as
> discussed near the top of gnome-print-pdf.c.

yes, we need to improve the encoding rutines.

> 
> I'll then create a compile-time option to encode images with
> RLE+ASCII85 instead of ASCIIHex.

I don't think we want to do it with a compile time function.
I think this should be a setting per device. It does not matter
how we implement it in the mean time, setting it on or off
is easy to do later. I think PS level 1 does not allow for 
RLE+ASCII85  but not sure about it. Anyone knows what PS level
is needed for this ?

> 
> My current implementation will add about 700 bytes of overhead to
> every postscript output file, to define a wrapper for showimage that
> makes decoding straightforward.  Is there any way around this for
> documents that don't use images at all?

Yes, if they don't use images we don't need to write the header.
This means that we will have to generate the postscript in two
passes. But again, lets implement it in a basci way and then we
can improve it. 700 bytes is not a huge deal too, specially
with the size gains that we get if we have images.
 
> I'll probably rewrite the current hex conversion stuff to use a hex
> encoder in the same style as the RLE and ASCII85 encoders, for
> consistency's sake.

makes sense.

> Anybody have any thoughts/advice/objections?

sounds great !

regards,
Chema