Re: [xsl] RE: XSL-FO and Z-index

Subject: Re: [xsl] RE: XSL-FO and Z-index
From: "Kevin Brown" <kevin@xxxxxxxxxxx>
Date: Sun, 23 Sep 2012 11:20:05 -0700
> Kevin can have it for speed, which I can't believe to be an issue for a
two page pdf.

Two pages, once? No, not really. Perhaps a tenth of a second. I would go
with what works for you *if a random two-page document once in a while is
your requirement*. 

I do not know what ultimate solution you are designing. I answered using a
solution that works for our customers (and by the way would also work for
your application if it is only a random two page document once in a while).
We have customers formatting 2,000,000 two page documents (like your phone
bill) for a monthly run, to be completed in a four hours. Add one tenth of
one second 2,000,000 times and you have added 55 hours of processing time
and would require four additional machines to complete formatting in the
time needed.

Also if you were producing print spool output, an image of a table grid
would be in a 100,000 page print spool once (a repeated image is referenced
on all pages, and included in the output file once). It is staged by the
printer and ready to print on all pages required. A print file using a table
grid drawn using borders would be 4 polygons per cell * number of cells on
the page * 100,000 pages. It would make the print spool file massive and the
printer would run very slow, if at all. 4 cells/row, 25 rows would be
10,000,000 draw instructions in a Postscript file. An image would be
essentially 100,001 instructions, one to store the image and 100,000 a
simple get to the stored image.

There are many ways to solve issues, one needs to take into account the full
requirements. If you do not care about performance in the sense given above,
then Ken's solution is fine. But if you are producing output destined for
print in large volumes, that approach will not work and your selected
solution would be "dead".

Kevin Brown
RenderX

Current Thread