Bug#712015: cups-filters: cups-pdf produces ugly pdfs through its pixelated font

[王晓林 <wx672ster@gmail.com>]

Hi, Brian

You guys are great! After changing 66 to 65, and a "sudo /etc/init.d/cups restart", the output PDF looks beautiful again! And the output PDF file name also gets friendly. When it's 66, the output file always named _stdin_.pdf overwriting previously generated one. Now it's 65, the output PDF file is named after the original source file, i.e. processing tmp.c gets a tmp.pdf.

Thanks again, Brian and Bastien, for your great help! Hope you can solve this bug fundamentally soon. If any more tests I can do, let me know.

Attached are PDFs generated with cost factors 66 and 65. You can see the deference.

Xiaolin

--

王晓林

On Fri, Jun 14, 2013 at 7:14 PM, Brian Potkin <claremont102@gmail.com> wrote:

On Thu 13 Jun 2013 at 17:22:20 +0800, 王晓林 wrote:

[Snipped: A very useful account of a testing procedure]

> In a short word, all the PS in /var/spool/cups look very good; all the PDF
> in ~/PDF look ugly. In the attachment, you can find
>

>    1. d00107-001, the PS file in /var/spool/cups
>    2. d00107-001.pdf, ps2pdf output
>    3. d00109-001.pdf, the error_log

This error_log (and the other one posted) show the PostScript file
being filtered as follows:

   PS in ---> pstopdf ---> pdftopdf ---> pdftops ---> then to cups-pdf

This is the PDF centric work flow. In principle there is nothing wrong
with this chain being used.

In practice an exception is made when the input is PostScript and the
output is being sent to a PostScript printer. The *cupsfilter line in
cups-pdf's PPD is commented out, so it is seen as a PostScript printer.

The filter chain is then:

   PS in ---> pstops---> then to cups-pdf

and this filtering sequence is used if /usr/share/cups/mime/mime.convs
contains

   application/postscript      application/vnd.cups-postscript 65 pstops

In cups 1.5.x the cost factor for pstops is indeed 65. In cups 1.6.2 the
cost factor is 66 (bug?).

The immediate solution to your 'bad' PDFs is to alter '66' to '65'.
However, I do not think this addresses the fundamental bug. Xiaolin,
would you please do some more testing? Basically, would you comment on
the appearance of the PDFs when text and PDF documents are sent to
cups-pdf with cost factors of 65 and 66?

Regards,

Brian.

Attachment: _stdin_.pdf
Description: Adobe PDF document

Attachment: tmp.pdf
Description: Adobe PDF document