Bug 691099 - Indeterminism in pdfwrite since pcl_fonts_mod merge
Summary: Indeterminism in pdfwrite since pcl_fonts_mod merge
Status: RESOLVED FIXED
Alias: None
Product: GhostPCL
Classification: Unclassified
Component: PDF Writer (show other bugs)
Version: master
Hardware: Macintosh MacOS X
: P4 normal
Assignee: Ken Sharp
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2010-02-02 12:30 UTC by Marcos H. Woehrmann
Modified: 2016-02-18 05:32 UTC (History)
0 users

See Also:
Customer:
Word Size: ---


Attachments
valgrind.log (29.46 KB, text/plain)
2010-02-02 12:33 UTC, Marcos H. Woehrmann
Details
screenshot.png (38.05 KB, image/png)
2010-02-04 10:24 UTC, Marcos H. Woehrmann
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Marcos H. Woehrmann 2010-02-02 12:30:05 UTC
Starting with r10645 the following command shows an indeterminism on peeves:

./main/obj/pcl6 -sOutputFile=24-01.pdf -sDEVICE=pdfwrite -r75 -dNOPAUSE 24-01.BIN
Comment 1 Marcos H. Woehrmann 2010-02-02 12:30:54 UTC
Created attachment 5934 [details]
24-01.BIN
Comment 2 Marcos H. Woehrmann 2010-02-02 12:33:25 UTC
Created attachment 5935 [details]
valgrind.log

Valgrind log for command line given in comment #0; I don't believe any of the
errors are relevant.
Comment 3 Marcos H. Woehrmann 2010-02-04 10:24:49 UTC
Created attachment 5943 [details]
screenshot.png

The attached screenshot shows the differences between runs with r10645.  This
is a zoomed in area of page 10, the only page that shows a difference.
Comment 4 Ken Sharp 2016-02-17 01:32:02 UTC
I can't reproduce this, and in fact I can't produce output that looks like the screenshot either (the screenshot looks wrong to me, both of the outputs). I suspect this was fixed by commit 7086f246fa047bcb9f3e9cfaacfbdae709058b91 which 'fixes' indexed (palette) colours in PCL.

http://git.ghostscript.com/?p=ghostpdl.git;a=commit;h=7086f246fa047bcb9f3e9cfaacfbdae709058b91

Marcos is there any way you can check this ?
Comment 5 Ken Sharp 2016-02-18 04:08:29 UTC
The problem initially reported 'seems' to be fixed, but there is a further problem with the 'default palette (B/W)' tests. Running that here on Windows always results in correct output, but the cluster tests result in coloured output, which is clearly wrong even without non-deterministic behaviour.

I'll try to reproduce that on Linux.
Comment 6 Marcos H. Woehrmann 2016-02-18 05:32:13 UTC
According to the nightly regression logs this issue was resolved on 2013-05-03 with a commit 9cd48ce016f8ce490d32679dab5de3cd781f6f85,