[gs-devel] Re: Merger ESP/GPL GS: GPL GS 8.57 is out, how to proceed with the merger

[Ray Johnston]

Ralph Giles wrote:
>> I have seen that GPL GhostScript 8,57 has been released. How should I 
>> porceed now with the merger? Should the branch now be merged into HEAD 
>> or are there any preparative steps needed?
>>     
>
> Ok, I don't immediately see a single strategy that would work for 
> both what you say CUPS needs and preserve the existing functionality
> for those who depend on it, so please proceed by making the core changes 
> depend on a switch, preserving the current behavior as the default. 
> Please do this on the branch, and then merge down to trunk after we've 
> verified it works on win32.
>   
And that it builds/works on Mac OS/X, and (if we still support builds on 
Mac Classic, verify that)
and run the full regression. Then merge.

Just to be entirely clear about the gs_init.ps change to default BG and 
UCR, changing these
defaults will affect the default appearance on CMYK devices when the 
colors are RGB.
While these values are easily controlled by customers/user with the 
corresponding PS
operators, most people don't set them. I think it is up to CUPS sending 
data to GS to
set these (then the defaults don't matter), or at least for gs to use 
different defaults when
running from CUPS.

Also the gs_setpd.ps change to Policy should default to PolicyNotFound 1 
and PageSize 0
(not 7). This is expected by some PS files and would change the results 
for some
users/customers that rely on the existing values. Once again, most 
people SHOULD set
these, and there is no problem with CUPS sending the setpagedevice the 
set these
Policy values and not rely on specific defaults.

CUPS doesn't really need to modify the default behavior, or to set any 
'option' switches
if it sends PS that sets the desired values using PS 
(setundercolorremoval, setblackgeneration
and setpagedevice). These changes to CUPS would still work with old 
(7.07?) gs.

While an option isn't needed, I don't object if that's the way you and 
Mike want to
proceed.

I don't have any problem with the addition of NOMEDIAATTRIBUTES, 
although I the
comment isn't clear to me:
< % Define NOMEDIAATTRS to turn off the default (but unimplementable) media
< % selection policies for setpagedevice.  This is used by CUPS to support
< % the standard Adobe media attributes.

Artifex has run the complete CET and has gone to some lengths to make 
the (complex)
setpagedevice media matching logic work like Adobe PS.

BTW, I disagree with calling the values of BG and UCR as a 'bug' as 
Martin Lottermoser
wrote. This is a matter of preference, and we chose this to avoid a 
problem with images.
Combining black ink with CMY causes color shifts and causes too much ink 
on inkjets
(which are the dominant color printers today). OTOH, using CMY for text 
can also be
a problem -- that's why 'real world' printers use different color 
mapping for text, graphics
and images. Look for 'switchable CRD' technology in gs later that will 
facilitate this.

I'll continue looking at the merge and let you know if I find anything 
else. I should have
some dead time while at WinHEC next week (or during our staff meeting on 
Thurs).

Regards,
Ray
------------------------------------------------------------------------