[reportlab-users] Is ReportLab suitable for this task?

[Jerome Alet]

On Fri, Jan 23, 2004 at 09:39:49AM +0100, Jerome Alet wrote:
> On Fri, Jan 23, 2004 at 08:24:32AM -0000, Andy Robinson wrote:
> >        
> > >     Processor : PII 350 Mhz
> > >     RAM : 288 Mb
> > >     Operating System : GNU/Linux
> > 
> > > the PDF generation part only took 216 seconds, i.e. 62 pages/sec
> > 
> > So a modern 2Ghz machine is likely to get 300-400 pages per second
> > Wow!  Maybe we could publish a case study on this?

it's there :

  http://www.librelogiciel.com/software/rlbench.tgz
  
WARNING : I've replaced the original logo with ReportLab Diagra's one,
          hoping you won't mind.
          
Code's license is ReportLab's one.

Read the comments at the top, then launch :

        python rlbench.py 20000
        
To create a 20000 pages document.        

It will print statistics on standard output.

WARNING : this eats a lot of memory, my own machine swaps and 
performance diminish. With less pages, and no swapping, I've 
got 63 pages / second on a PII 350.
          
On my* Bi-PIV Xeon 2.2 Ghz, I can generate 100000 pages in 340 
seconds (293 pages / second) and the resulting file is 85 Mb long
(WHO wants to print that ?). I guess it could do more, but I'm
short on memory and the machine is already under some load.
Memory consumption is more than 720 Mb just before the end, i.e.
RL's structures take 7.5 Kb per page in memory, and 880 bytes per
page on disk.

*: no it's not mine, unfortunately...

hoping this helps

bye

Jerome Alet
-- 
"A non-free program is a predatory social system that keeps people 
in a state of domination and division, and uses the spoils to 
dominate more." - RMS