[cups.general] Sense and nonsense of RIPCache

[Till Kamppeter]

On 07/27/2010 04:49 PM, Tim Waugh wrote:
> On Tue, 2010-07-27 at 10:16 -0400, Gene Heskett wrote:
>> It appears some distros are already doing this Till, I cannot grep for
>> RIPCache in my /etc/cups directory.  And cups is reported to be:
>
> Gene, this just means that the CUPS configuration doesn't set a
> different value than the default that the ghostscript 'cups' driver is
> compiled to use.  The use of RIPCache is hard-coded into the ghostscript
> 'cups' driver, i.e. the driver that produces output in CUPS Raster
> format.
>
> Till, I've seen this same problem the other day, and no value of
> RIPCache would work around it.  Sounds like using automatic allocation
> in ghostscript is the solution...

I think then I will modify the CUPS Raster driver in Ghostscript so that 
it lets Ghostscript choose the cache memory sizes if the value of 
RIP_MAX_CACHE is not valid and recommend to distributions to default 
RIPCache to be set to "auto". This will make stuff "just work" but still 
give to system admins the possibilities which they had before.

Or, as manual RIP_MAX_CACHE has always the risk of failure and letting 
Ghostscript decide always works and even with reasonable memory 
consumption, we should perhaps even better remove the RIP_MAX_CACHE 
support from Ghostscript's "cups" output device completely. WDYT?

    Till