Re: [gnome-print] Filter code for gnome-print

[Scott Gifford <sgifford suspectclass com>]

Chema Celorio <chema@ximian.com> writes:

> On Wed, 2002-06-05 at 22:03, Scott Gifford wrote:
> > Chema Celorio <chema@ximian.com> writes:
> > 
> > > On Thu, 2002-05-09 at 16:49, Scott Gifford wrote:
> > > ..snip..
> > > >   I've been thinking about this, too.  I'm not sure where the code
> > > > went from my Web site, but I'll pull it out tonight or this weekend
> > > > and post a new link to it.  I'd love to help get it integrated.
> > > 
> > > Great!
> > 
> > OK, I've dug out what I had before, and it's available at:
> > 
> >     http://www.suspectclass.com/~sgifford/filter/

[...]

> > Who should I coordinate CVS commits with?  Should I post patches here
> > first, and get an OK, or just commit when I think things are ready?
> 
> Myself please. You can send me patches for the initial commit.

Looks like my original post wasn't accepted because the patch and
tarfile were too big, so here it is again with URLs.

You can get the patch and tarfile referenced below at:

    http://www.suspectclass.com/~sgifford/filter/files/gnome-print-filter3.patch 
    http://www.suspectclass.com/~sgifford/filter/files/gnome-print-filter3.tar.gz 

Anyways...

OK, here you go.  It's been tested a bit, and seems to work fine.

First, there's a patch that applies against tonight's CVS.

Second, there's a tarfile with the regression tests for the filter
library, which contained some binary files so didn't work well with
diff.

Some files are in both the patch and the tarfile; they're identical,
so either one is fine.  I'd re-do the patch to avoid this redundancy,
but I just tested it pretty thorougly, and would rather leave well
enough alone.

The patch enables RLE+ASCII85 encoding for all images, with an
infrastructure for user-defined and flexible encoding that I think
we'll be able to make use of later on.

It adds a small test to the generate program when --sequence=image.

It adds a README.filters file, which is partly babbling garbage, but
contains some good stuff at the end.  I'll clean this up later.

The filters come with a (perhaps excessively) thorough set of
regression tests.  Go into tests/filter and run ./test to encode 11
different files with 20 different combinations of encoding and 7
different blocksizes (I told you it was excessive...), then decode
them with ghostscript and make sure that they're identical to the
original files.  All of these tests pass on my machine; if any of them
fail, something's dreadfully wrong.

Images are now handled by a postscript gnomeprintimage command, which
is included in the prologue IFF any images are used in the document.

Please let me know what questions you have, and what needs to be
fixed!

-----ScottG.