Created attachment 2920 [details]
File that demonstrates incorrect tiffsep behavior

Comment on attachment 2920 [details]
File that demonstrates incorrect tiffsep behavior

Private customer data

gsc -sDEVICE=tiffsep -dBATCH -dNOPAUSE -sOutputFile=./temp -r20 -f 20699JFP301.pdf

head causes all separations to show up, however file exhibits other unexpected
behavior possibly due to an object on the not-reported separation of /All.

I ran the job as described and see 5 spot plates, but output doesn't look
correct to me.

To come at it another way I tried the job with these parameters:

-q -sDEVICE=psdcmyk -dBATCH -dNOPAUSE -sOutputFile=20699JFP301-%d.tif -r20 -f
~/20699JFP301.pdf 


and gs segfaults in pdf14_cmykspot_put_image:

Program received signal EXC_BAD_ACCESS, Could not access memory.
Reason: KERN_INVALID_ADDRESS at address: 0x02b9d784
0x00107900 in pdf14_cmykspot_put_image (dev=0x0, pis=0x2bfd83b,
target=0x300f034) at ./src/gdevp14.c:1410
1410                        comp = buf_ptr[x + planestride * input_map[comp_num]];

Note the null dev pointer.

There is a bug here having to do with the fact that the function pdf14_open
needs to be made aware of the number of spot colors in the job in order to tell
pdf14_ctx_new the correct number of color components so that the buffer can be
sized correctly down in pdf14_buf_new.

I temporarily changed pdf14_buf_new to force the correct number of components
specific to this particular job. I did this to see what the output would look
like with a properly sized output buffer. When I did this, the output from gs
matched the output from Acrobat and Apple Preview. As in the gs output, for
Acrobat and Prevew the colors "186", "bluedie" and "Varnish" knock-out
everything else in the document. These correspond to gs output plates "s1", "s3"
and "s5", which all appear as exptected given this knockout behavior.

Since the fundemantal problem of calculating the correct number of components
for sizing the buffer correctly has already been noted in bug report #688778,
this bug is being closed as duplicate of #688778. Please refer to that bug
report to track the fix for the buffer size issue.


*** This bug has been marked as a duplicate of 688778 ***