Re: Ghostscript/GhostPDL 9.28 Release Candidate 1

Chris Liddell <[email protected]> Tue, 20 Aug 2019 07:53:30 +0100
Newsgroups gmane.comp.printing.ghostscript.devel
Message-ID <[email protected]>
On 20/08/2019 07:11, Akira Kakuto wrote:
> Dear Chris,
> 
>> I suspect that the new code fails to read iccprofiles in the
>> SAFER case if compiled with COMPILE_INITS=0, and setpagedevice is
>> contained.
>> In the case of usual eps without setpagedevice,
>> ps2pdf foo.eps foo.pdf
>> succeeds without any problem, so I believe that iccprofiles are read
>> successfully in these cases.
> 
> I confirmed that if I compile with COMPILE_INITS=1,
> all my problems disappear:
> ps2pdf iii.eps iii.pdf
> that is, -dSAFER works fine with setpagedevice.
> 
> Please check the case COMPILE_INITS=0.
> In my case, gs-9.28rc1 fails to read iccprofiles in -dSAFER case
> if setpagedevice is found. The iccprofiles can be read in
> -dSAFER -dOLDSAFER and, of course, -dNOSAFER cases.

Hmmm, that is odd... I definitely add the ICC profiles path to the
reading permitted list. I wonder if there's a subtlety about it that
I've missed. It may be that the path I add is only correct when using a
command line option to explicitly set the ICC profiles path.

I'll take a look.

Chris