If you can find a way to create a minimal script, that would be better,
because then we can verify the fix.
But barring anything else, seeing the .eps might help.
Michael Hearne wrote:
> John - Thanks for the quick fix. I'm having a separate issue now with
> an EPS file being generated (using savefig() again) that appears to be
> invalid (can't display it in OS X Preview, or convert to PDF with
> ps2pdf). Unfortunately, the code that creates this particular file is
> rather involved, and hard to replicate with one simple script.
> If I send the offending EPS file (it's 1.4 MB zipped), is it possible to
> tell from that what the problem is?
> The errors from ps2pdf are:
> Error: /undefined in --get--
> Operand stack:
> true --dict:43/43(ro)(L)-- .notdef
> Execution stack:
> %interp_exit .runexec2 --nostringval-- --nostringval--
> --nostringval-- 2 %stopped_push --nostringval--
> --nostringval-- --nostringval-- false 1 %stopped_push 1905
> 1 3 %oparray_pop 1904 1 3 %oparray_pop --nostringval--
> 1888 1 3 %oparray_pop 1771 1 3 %oparray_pop
> --nostringval-- %errorexec_pop .runexec2 --nostringval--
> --nostringval-- --nostringval-- 2 %stopped_push
> --nostringval-- %finish_show --nostringval-- --nostringval-- 5
> 4 0 --nostringval-- (pdf_text_enum_t) %op_show_continue
> Dictionary stack:
> --dict:1144/1684(ro)(G)-- --dict:0/20(G)-- --dict:75/200(L)--
> --dict:61/112(L)-- --dict:20/25(ro)(L)--
> Current allocation mode is local
> Last OS error: 2
> Current file position is 996122
> GPL Ghostscript 8.61: Unrecoverable error, exit code 1
> John Hunter wrote:
>> On Wed, Jun 11, 2008 at 10:14 AM, John Hunter <jdh2358@...> wrote:
>>> Interesting. When I plot it on my screen it looks correct with a *Agg
>>> GUI backend. But when I save it (either from the GUI or using
>>> savefig) it has the problem you describe. This suggests to me that
>>> either some cached information or something is not getting notified of
>>> the different default dpi in savefig. I'll look into it. Since PS is
>>> dpi independent, it would not affect it.
>> Well, my diagnosis wasn't correct, but I fixed the problem. In the
>> caching of the text layout, we were not taking into account the
>> different renderers in the cache key. Since PS has a nominal dpi
>> setting that is ignored by postscript, the differences in dpi were not
>> getting picked up in the layout cache. The solution I implemented was
>> to simply add the renderer id to the property tuple used for caching.
>> Fixed in svn r5470.
Science Software Branch
Operations and Engineering Division
Space Telescope Science Institute
Operated by AURA for NASA