Re: Re: [gs-cvs] gs/lib

Igor,

Igor V. Melichev wrote:
> 
> Now I've tested the patch and detected a regression with 031-05.ps .
> The generated PDF file contains
> 
> q
> 0 0 m
> W n
> q
> 10 0 0 10 0 0 cm BT
> /R12 10.9 Tf
> 1 0 0 1 202.124 178 Tm
> (Genoa Technology Genoa Technology )Tj
> ET
> Q
> Q
> 
> and the text must not be painted.
> 
> I would very appreciate if you test patches before commitment.

I did test the patch with all of the PDF files in the reqression suite,
just not with the PDF files created by our pdfwrite from Genoa tests.

I've traced the problem to a strange "feature" in the pdf_ops.ps.
It specifically uses 'fsexec' to preserve currentpoint across the
fill/stroke operations. The reason for this is not documented and
I've traced this to a change made between versions 3.45 and 3.46.
The only NEWS entry that seems to relate to this is:

------------
Fixes bugs:
         - PDF to PostScript conversion didn't work.  (pdf_main.ps)

Shifts most of the work in the PDF interpreter from the reader to the
ProcSet.  This simplifies the code considerably and also makes it easier to
modify.  (gs_pdf.ps, pdf_draw.ps, pdf_font.ps, pdf_main.ps)
------------

I will be committing a change to remove the saving of 'currentpoint'
from fsexec once my local regression testing completes, then I will
also change the '/n { }' definitions to use on 'currentpoint'
which will be Adobe Acrobat reader compatible.

Regards,
Ray
______________________________________________________________________