xdvi figure rendering broken with gs 9.26

[Paul Vojta]

Indeed I am looking at it.  Using "git bisect" I found that the responsible
commit is 533f78554f194524abb55183d370e64354caa329 (in the ghostscript git).

That commit removed the definition of /execute, which xdvi definitely uses.

I'll look at it more this weekend.

Now there's still a bug with fragments of a figure appearing in subsequent
renderings via gs.  I haven't looked at that yet.

But, the current use of gs was modeled after ghostview, which has since
transitioned to using libgs.  Since xdvi's gs code is a bit of a hack,
it too would benefit from switching to libgs.

What do you think of xdvi using libgs sometime down the road?

Paul


On Tue, Jul 09, 2019 at 04:49:09PM -0600, Karl Berry wrote:
> Hi,
> 
>     https://bugs.mageia.org/show_bug.cgi?id=24941
> 
> FYI, xdvi(k) is primarily maintained by its original author,
> Paul Vojta (cc'd). Upstream is http://xdvi.sourceforge.net/ ...
> 
>     xdvi figure rendering broken with gs 9.26
> 
> Your example document and figure work for me with xdvi (from tl'19) and
> gs-9.26.  It's with gs-9.27 that the figure is not displayed.
> 
> This is almost certainly because gs-9.27 (which was released as we were
> finalizing TL'19, so there was no chance to deal with their induced
> incompatibilities) removed many operators.
> https://www.ghostscript.com/doc/9.27/History9.htm#Version9.27
> 
> I hope Paul will be able to look at it.
> 
> Thanks for the report. --karl