Summary: | bbox device returning different values | ||
---|---|---|---|
Product: | Ghostscript | Reporter: | evan.cooch |
Component: | Font API | Assignee: | Chris Liddell (chrisl) <chris.liddell> |
Status: | RESOLVED DUPLICATE | ||
Severity: | normal | CC: | chris.liddell |
Priority: | P4 | ||
Version: | master | ||
Hardware: | PC | ||
OS: | Windows 7 | ||
Customer: | Word Size: | --- | |
Attachments: |
MWE
.bat file executing GS commands... PDF of MWE PNG file using bbox generatred using gs9.27 DVI from MWE |
Description
evan.cooch
2019-04-09 02:25:25 UTC
Created attachment 17310 [details]
MWE
The MWE referred to in bug report.
Created attachment 17311 [details]
.bat file executing GS commands...
/bat file to execute commands to generate bbox for equation in MWE.
Just rolled back through gs9.27 -> 9.25. The bug emerges between gs9.26 and gs9.27. Here are the bbox numbers for 9.25 %%BoundingBox: 262 632 348 657 %%HiResBoundingBox: 262.379101 632.411981 347.957989 656.243980 9.26 %%BoundingBox: 262 632 348 657 %%HiResBoundingBox: 262.379101 632.411981 347.957989 656.243980 9.27 %%BoundingBox: 258 629 343 676 %%HiResBoundingBox: 258.047992 629.171981 342.431990 675.539979 So, you can se the way gs9.27 is handling/calculating the bbox is different 9broken), relative to earlier releases. I'm only seeing this for a simple problem, but my guess is this ha all sorts of ramifications for other things using pdflatex. Please attach a Postscript or PDF file that illustrates the problem. Tis clearly isn't anythign to do with images. Most likely its a change in the bbox device but I'll categorise it as PDF interpreter for now. Created attachment 17313 [details]
PDF of MWE
As requested - PDF of the MWE as described in original report.
Created attachment 17314 [details]
PNG file using bbox generatred using gs9.27
PNG showing right-truncation of equaition, because bbox generated by 9.27 is 'shifted left (and up, although it doesn't show the latter here), relative to 9.26, 9.25,...
Created attachment 17315 [details]
DVI from MWE
If instead of pdflatex, I use latex - DVI, I generate the attached .dvi file. If I then use
"dvipng" -q -D 1200 -T tight -bg Transparent -o "mwe_latex-dvi.png" "mwe_latex-dvi.dvi"
the PNG file is generated perfectly. Problem only arises is someone is using pdflatex -- and gs9.26 -- to get there from here. pdflatex works fine with all earlier versions of gs (that I tried, at least).
I'm afraid bisecting reveals this infamous commit as the source of the problem: http://git.ghostscript.com/?p=ghostpdl.git;a=commit;h=355434f4b1bbe8c4f98cafad5a6868aa2f0eaae1 (In reply to evan.cooch from comment #8) > Created attachment 17315 [details] > DVI from MWE > > If instead of pdflatex, I use latex - DVI, I generate the attached .dvi > file. If I then use > > "dvipng" -q -D 1200 -T tight -bg Transparent -o "mwe_latex-dvi.png" > "mwe_latex-dvi.dvi" > > the PNG file is generated perfectly. Problem only arises is someone is using > pdflatex -- and gs9.26 -- to get there from here. pdflatex works fine with > all earlier versions of gs (that I tried, at least). I meant.. "Problem only arises is someone is using pdflatex -- and gs9.27 -- to get there from here. pdflatex works fine with all earlier versions of gs (that I tried, at least)." gs9.26 works fine regardless of how the .PDF is compiled (latex or pdflatex). *** This bug has been marked as a duplicate of bug 700952 *** |