RE: mxe-octave: ghostscript 9.07, why not 9.10

[Серёжа Плотников]

31.07.2014 22:05 пользователь "John D" <address@hidden> написал:
>
>  
>
>  
>
> From: John D [mailto:address@hidden]
> Sent: Thursday, July 31, 2014 3:22 PM
>
> To: 'sergey plotnikov'; address@hidden
> Subject: RE: mxe-octave: ghostscript 9.07, why not 9.10
>
>  
>
>  
>
>  
>
> From: John D [mailto:address@hidden]
> Sent: Wednesday, July 30, 2014 2:29 PM
> To: 'sergey plotnikov'; address@hidden
> Subject: RE: mxe-octave: ghostscript 9.07, why not 9.10
>
>  
>
>  
>
>  
>
> From: sergey plotnikov [mailto:address@hidden]
> Sent: Wednesday, July 30, 2014 1:22 PM
> To: John D; address@hidden
> Subject: Re: mxe-octave: ghostscript 9.07, why not 9.10
>
>  
>
> On 29/07/14 22:45, John D wrote:
>>
>>  
>>
>>  
>>
>> -----Original Message-----
>>
>> From: John D [mailto:address@hidden]
>>
>> Sent: Tuesday, July 29, 2014 4:40 PM
>>
>> To: address@hidden
>>
>> Cc: 'sergey plotnikov'
>>
>> Subject: mxe-octave: ghostscript 9.07, why not 9.10
>>
>>  
>>
>>  
>>
>> Message: 6
>>
>> Date: Tue, 29 Jul 2014 10:09:19 +0200
>>
>> From: sergey plotnikov <address@hidden>
>>
>> To: address@hidden
>>
>> Subject: mxe-octave: ghostscript 9.07, why not 9.10
>>
>> Message-ID: <address@hidden>
>>
>> Content-Type: text/plain; charset="iso-8859-1"; Format="flowed"
>>
>>  
>>
>> Hi all,
>>
>>  
>>
>> I've cross-built recent development version of Octave for Windows using
>>
>> MXE-Octave (also, one of the recent csets) and after some testing get into a
>>
>> trouble which I'm not sure caused by Octave itself. Sometimes when trying to
>>
>> print a figure to pdf (print('-dpdf',..)) Octave may hang on waiting before
>>
>> gs.exe completed writing into a pdf-file. But as a matter of fact gs.exe
>>
>> doesn't want to finish and continues writing into pdf inflating its size
>>
>> gradually approaching infinity (last time I forgot to kill gs.exe process my
>>
>> one-page pdf reached 600Gb).
>>
>>  
>>
>> This bug looks quite similar to GS bug #693591
>>
>> [http://bugs.ghostscript.com/show_bug.cgi?id=693591] which was fixed in
>>
>> gs9.10. And according to my recent testing, use of separately installed
>>
>> gs9.10 instead of default Octave gs.exe helps.
>>
>>  
>>
>> So my questions is rather easy, do you by any chance plan to switch GS
>>
>> version in MXE-Octave from 9.07 to 9.10? If not, it might worth to apply a
>>
>> patch fixing this bug in GS source
>>
>> [http://git.ghostscript.com/?p=ghostpdl.git;a=commit;h=dfd7fc9fe152245256e43
>>
>> 7141dd7780c8313c971].
>>
>>  
>>
>> Sorry, if I overlooked existing discussion on this topic.
>>
>>  
>>
>> Sergey
>>
>>  
>>
>> ----
>>
>>  
>>
>> Why not 9.14 ?
>>
>> Patch works for me under native mingw and cross compiled on my fedora
>>
>> machine
>>
>>  
>>
>> --
>>
>>  
>>
>> It of course doesn't fix the issues with octave pdfwrite tho
>>
>>  
>
> Sorry, what issues do you mean?
>
> I've tried your patch today. Octave has been successfully cross-compiled, but simple printing to pdf has crashed with the error: "GPL Ghostscript 9.14: ERROR -12 ...", which is according to psi/ierrors.h is io error from the block of PS level 1 errors.
>
> Interestingly, if I use gs9.14 downloaded from official gs web-site (win32 version) the same printing to pdf works just fine.
>
> Sergey
>
>  
>
> -
>
> Well I thought the issue was just as per  41141 but I just realized that octave found another version of ghostscript and was using that.
>
> I am building a native version so will see if that works.
>
>  
>
> --
>
>  
>
> Running a native built version of ghostscript works for creating a pdf, but complains for postscript:
>
> Unknown device: p
>
> Unrecoverable error: undefined in .uninstallpagedevice
>
> Operand stack:
>
>     Defaultdevice
>
>  
>
> So I will have to take a look at the cross built version and see what if different with it.
>
>  
>
> --
>
>  
>
> Looking at it more, my cross built ghostscript 14 does create pdfs ok so is behaving like my native version.
>
> What OS you cross compile on?

That's weird.. I'm cross-compiling on Ubuntu 13.10 x64, using freshly cloned mxe-octave repo with your patch applied to the latest cset.

Sergey