GS 9.51 and transparency

[Bruno Voisin]

>From the MacTeX team, answering Uwe Siart's message on transparency in gs 9.51: I'm not sure that's the same issue, but the consequences of the new implementation of -dSAFER in gs 9.50 on transparency in PSTricks was discussed on the MacTeX list and directly with Chris Liddell (of the Ghostscript dev team) in late October last year.

As a result he implemented a new -dALLOWPSTRANSPARENCY switch, bringing back transparency while still in -dSAFER mode:

http://git.ghostscript.com/?p=ghostpdl.git;a=commit;h=d1eac80695b2c6f2a85942a82518715549541969

The problem seems to be that PSTricks implements transparency for distiller with "/SetTransparency pdfmark" which I imagine is standard (I'm no PS/PDF expert), but for gs it uses .setopacityalpha which is documented by gs as non-standard:

https://www.ghostscript.com/doc/current/Language.htm#Additional_operators

This link also describes the new -dALLOWPSTRANSPARENCY switch.

(The following based on my recollection of correspondence with Karl back then; I'll let Karl correct anything that needs correcting.)

We then evoked with Karl the idea to modify epspdf, epstopdf and simpdftex such that -dALLOWTRANSPARENCY is called by default.

Karl wasn't fond of the idea, for two reasons mostly: first, if -dALLOWTRANSPARENCY were really safe, it would be the default; and second, that would only solve the problem for gs 9.51 and above.

So the conclusion was to leave things as they were and instead to warn the user that gs needed to be called with -dALLOWPSTRANSPARENCY if PSTricks transparency was looked for.

Back then I had ended my last message on the topic telling I would contact Herbert Voss about the matter, but then forgot and never did. Sorry!

Is there a test file that could be used to test this (I'm no pstricks user myself)?

Bruno Voisin