Picon

Ghostscript issue when parallel processing

Good afternoon everyone,

I have a problem because I find two different behaviours in two servers using GhostScript 9.14

The first server has 2x2GHz processors (12 physical cores, 24 logical) 144 GB RAM

The second has 4x2.3GHz processors (48 physical cores, 96 logical) 512 GB RAM

Both are running a Windows Server 2012 R2.

When I proceed to try sequencial processing (one by one) the second processor is faster, as it is expected. But when we try parallel processing (several threads at same time) the second processor deteriorates more (in time) than the first. As much open threads, much is the difference between the first and the second.

Anyone has some idea of what can be happening?

Thanks all in advance,

Best Regards,

Héctor Martínez

Ken Sharp | 6 Jun 15:56 2016

Re: PostScript to PDF and Duplex/Tumble

At 13:52 06/06/2016 +0000, RITTER, Philippe wrote:

>I used this flag also, but I don't di dit in all my GS call.
>
>After review, I made it now each time, and it works !

Ah great, good to know it works

                     Ken

Ken Sharp | 6 Jun 15:39 2016

Re: PostScript to PDF and Duplex/Tumble

At 13:08 06/06/2016 +0000, RITTER, Philippe wrote:

>I have also tried to rotate with your sample. It works in Ghostview, but 
>not when I try to convert it with gs to PDF. Same for /Orientation, 
>doesn't work either. It was my typo error, in my postscript it was correct.

The pdfwrite device has /AutoRotatePages set to /PageByPage by default, 
which will rotate pages so that the text is upright and horizontal, you'll 
need to set the to /None as well when using pdfwrite.

                         Ken

Ken Sharp | 6 Jun 10:53 2016

Re: PostScript to PDF and Duplex/Tumble

Hi Philippe,

Sorry you've had to wait for a reply, we've had a company staff meeting and 
so we've all been away from our desks.

At 17:02 01/06/2016 +0000, RITTER, Philippe wrote:

>I'm using Ghostscript for converting a Postscript generated file to PDF. 
>It is working fine, except if in the postscript file there is this command :
><</Duplex true
>/Tumble true
>/EndOfSet false
> >>setpagedevice
>
>The Tumble is not recognized. I think it is maybe by design, because 
>Duplex is not possible in PDF.

None of those requests will have any effect with the pdfwrite device. 
Device-specific requests are ignored by devices which don't implement them; 
since Tumble is only relevant with Duplex, any device which does not 
implement Duplex will also not implement Tumble.

Because the PDF specification doesn't allow for print features like these, 
they cannot be embedded in the PDF file and so the pdfwrite device does 
indeed ignore them.

>  But as my document is full duplex, and printed on a Canon Inkjet system, 
> I want that in my PDF some pages are turned 180° for printing like Duplex 
> Tumbled.

Can you not simply specify Duplex and Tumble when you print the PDF file to 
your printer ?

>I have also tried to put /orientation 2, but it doesn't work either.

All the entries in the page device dictionary are *requests*, which a 
device may honour or ignore as it appropriate. Note that the correct key 
should be /Orientation, not /orientation, PostScript is case-sensitive.

>Can someone help me or give me some information how I could achieve this ? 
>Maybe in the Postscript file directly, or in command line in gs ?

You can rotate the pages if you really want to do so by providing a 
BeginPage procedure which does what you want. This procedure is called at 
the start of every page and is permitted to do operations that affect the 
CTM after the page is established but before the program is executed.

This code:

<<
/BeginPage
{               %% starts with number of showpages so far on the stack
   2 mod 0 eq    %% if <page> mod 2 is 0 then its an even page
   {             %% so we rotate it
     currentpagedevice /PageSize get
     aload pop -1 mul
     translate
     180 rotate
   } if
}
 >> setpagedevice

Will rotate even numbered pages by 180 degrees.

                 Regards,

                     Ken Sharp

RITTER, Philippe | 1 Jun 19:02 2016
Picon

PostScript to PDF and Duplex/Tumble

Hello Everyone,

This is my first message to this list, i'm looking for some help.

I'm using Ghostscript for converting a Postscript generated file to PDF. It is working fine, except if in
the postscript file there is this command :
<</Duplex true
/Tumble true
/EndOfSet false
>>setpagedevice

The Tumble is not recognized. I think it is maybe by design, because Duplex is not possible in PDF. But as my
document is full duplex, and printed on a Canon Inkjet system, I want that in my PDF some pages are turned
180° for printing like Duplex Tumbled.

I have also tried to put /orientation 2, but it doesn't work either.

Can someone help me or give me some information how I could achieve this ? Maybe in the Postscript file
directly, or in command line in gs ?

Thank you and regards
Philippe RITTER

Ulf Wittenberg | 31 May 16:04 2016
Picon

overprint simulation of spot colors - change from 9.16 to 9.19?

Dear list members,

today I tried my proven commands for rendering a CMYK TIFF (tiff32nc) from PDFs containing overprinting CMYK and spot color elements.
v9.16 creates a correct CMYK comp representation with overprint simulation of of both - v9.19 does not simulate the spot color overprints.
Following commands work flawlessly with 9.14 and 9.16:

/usr/local/bin/gs -dSAFER -dBATCH -dNOPAUSE -dNOPROMPT -dUseMediaBox -dAlignToPixels=0 -dGridFitTT=2 -dTextAlphaBits=4 -dGraphicsAlphaBits=4 -dSimulateOverprint=true -sICCProfilesDir="/opt/local/share/ghostscript/9.19/iccprofiles/" -dUsePDFX3Profile -dRenderIntent=1 -dBlackPtComp=1 -sNamedProfile="/Volumes/XYZ/GhostScript_ImageMagick_PDF2JPEG/spotcolors_lab_values.txt" -sDEVICE=tiff32nc -r200 -sOutputFile="/Volumes/XYZ/GhostScript_ImageMagick_PDF2JPEG/output_%03d.tif" "/Volumes/XYZ/GhostScript_ImageMagick_PDF2JPEG/input.pdf"

Just a note:
I am using my own named color table to get better spot-to-CMYK color conversions, entries look like this:
4357;
PANTONE 100 C, 92.040, -7.560, 65.760;

Screenshot of both results is attachment. If attachment should get scrubbed by list, here’s link including the source PDF: http://kurz.link/3hY

Does anybody have an explanation for this change in behavior?


Thanks in advance,
Ulf Wittenberg


terry | 13 Apr 22:36 2016

pxl "Staple" patch?

Sorry if this is a dupe. I sent it from the wrong account earlier.

********

Does anybody know when/if

"Bug 696314 [1] - Add PJL to staple documents": 
http://bugs.ghostscript.com/show_bug.cgi?id=696314

would appear in a release?

It enables a "staple" option for printers that have a stapler.

Hin-Tak Leung Uploaded a patch 
(http://bugs.ghostscript.com/attachment.cgi?id=12348&action=diff), but I 
can't figure out from Bugzilla or Git what happens next or where it 
would appear.

Thanks!

Terry

--

-- 
http://bupkis.org
If you're not here, you haven't got bupkis!
Chris Liddell | 30 Mar 19:45 2016

Re: Wrong architecture on 64 bit build?

genarch.exe creates arch.h, in this case, by copying
arch\windows-x64-msvc.h into obj\arch.h.

The best I can suggest here is to make sure that the obj and obj64
directories are deleted before trying again.

Chris

On 30 March 2016 at 14:55,  <terry <at> cnysupport.com> wrote:
> Well, it's different now. 8-)
>
> I took the command line from the VS "build" command and ran it in the 64 bit
> VS command window and got this:
>
> .\obj64\aux_\genarch.exe .\obj64\arch.h .\base\..\arch\windows-x64-msvc.h
> NMAKE : fatal error U1045: spawn failed : Invalid argument
>
> .\obj64\arch.h does indeed, not exist and I don't see it in any of the
> directories.
>
> Do you have any idea what creates it or where it's hiding?
>
> I appreciate the help. I'm just trying to make a tiny change, but it's being
> uncooperative.
>
> When I get this working, I'll submit a patch so "ARCHAIC" is a switch in the
> epson device, not a #define. It's used with some old Epson-compatible
> printers.
>
> Thanks,
>
> Terry
>
>
>
>
>
> **************************************************************
> * Assuming that target has SSE2 instructions available. If   *
> * this is NOT the case, define DONT_HAVE_SSE2 when building. *
> **************************************************************
>         if not exist .\obj64 mkdir .\obj64
>         if not exist .\obj64 mkdir .\obj64
>         if not exist .\obj64\cups mkdir .\obj64\cups
>         if not exist .\obj64 mkdir .\obj64
>         if not exist .\obj64 mkdir .\obj64
>         if not exist .\obj64 mkdir .\obj64
>         if not exist .\obj64 mkdir .\obj64
>         if not exist .\obj64\aux_ mkdir .\obj64\aux_
>         if not exist .\bin mkdir .\bin
>         echo     -W2 /wd4996 /wd4224 /nologo /MT -DCHECK_INTERRUPTS
> -D_Windows -D__WIN32__ > .\obj64
> \ccf32.tr
>         cl /wd4996 /wd4224       /DHAVE_SSE2   /FR.\obj64\  <at> .\obj64\ccf32.tr
> .\base\echogs.c -Fo.\ob
> j64\aux_\echogs.obj /Fe.\obj64\aux_\echogs.exe /link ""
> echogs.c
>         cl /wd4996 /wd4224       /DHAVE_SSE2   /FR.\obj64\  <at> .\obj64\ccf32.tr
> .\base\genconf.c /Fo.\o
> bj64\aux_\genconf.obj /Fe.\obj64\aux_\genconf.exe /link ""
> genconf.c
>         cl /wd4996 /wd4224       /DHAVE_SSE2   /FR.\obj64\  <at> .\obj64\ccf32.tr
> /Fo.\obj64\aux_\genarch
> .obj /Fe.\obj64\aux_\genarch.exe .\base\genarch.c /link ""
> genarch.c
>         .\obj64\aux_\genarch.exe .\obj64\arch.h
> .\base\..\arch\windows-x64-msvc.h
> NMAKE : fatal error U1045: spawn failed : Invalid argument
> Stop.
>
>
>
>
Terry Carmen | 30 Mar 02:18 2016

Wrong architecture on 64 bit build?

I compiled the latest GS source on 32 Bit Windows w/Visual Studio 2010, 
and targeted a 64 Bit environment.

Everything seemed to work, however I get a message on the 64 bit machine 
about it being a 32 bit DLL, even though it's named GSDLL64.DLL, and I 
get a load error (126?).

All I need to do is set the "ARCHAIC" switch in the epson device and 
recompile, but it's not cooperating.

Does anybody have any suggestions?

Thanks!

Terry
Hartmut Goebel | 25 Mar 15:30 2016

gsapi_set_stdio() and -OutputFile=-

HI,

I'm the developer of python-ghostscript, an interface to use the
ghostscript api from Python. ATM I try to complete a test-suite. And I'm
failing to redirect stdout using gsapi_set_stdIO(). Unfortunatly
gsapi_set_stdIo is not well documented.

The output-device I'm using is bmp16.

If passing -OutputFile=-, the image data is written to the stdout
instead of being passed to the stdout-function I've set with
gsapi_set_stdio().

And when leaving -Outputfile away, I get the error:
   GPL Ghostscript 9.14: Could not open the file .
   GPL Ghostscript 9.14: Unrecoverable error, exit code 1

On IRC, Robin Watts already answered:

> That doesn't entirely surprise me.

> This is *not* my area really, but I thought that the purpose of the
stdout redirection was so you could redirect the backchannel stuff from
postscript.

> When you use -OutputFile=- that's an instruction to the device, so a
different kettle of monkeys.

So is there some way to make the bmp16-device to use the stdout_fn?
Or which other device should I use?
If this does not help, I'd need some other test, eg. a small postscript
file writing "Hello World" to stdout. Any idea for this?

--

-- 
Regards
Hartmut Goebel

| Hartmut Goebel          | h.goebel <at> crazy-compilers.com               |
| www.crazy-compilers.com | compilers which you thought are impossible |
Chris Liddell | 23 Mar 15:37 2016

Re: Ghostscript and GhostPDL 9.19 release

I've added the files, but I was contemplating doing away with them,
since we moved to github as the download server.....

Anyone any thoughts on that?

Chris

On 23/03/16 14:31, Johannes Meixner wrote:
> 
> Hello Chris,
> 
> On Mar 23 11:38 Chris Liddell wrote (excerpt):
>> Get it here:
>>
>>    http://www.ghostscript.com/download
> 
> On
>   http://www.ghostscript.com/download/gsdnld.html
> near the bottom both links to
>   MD5 Checksums
>   SHA1 Checksums
> result 404 (page not found) errors.
> 
> 
> Kind Regards
> Johannes Meixner

Gmane