[reportlab-users] Report generation time is not optimized - ideas?

[J. R. Carroll]

Hi Tim,

Yeah, I'm aware that our OE platform is the main culprit of our slow PDF
generation (in fact, most people are surprised that on first pass the
reports are built in that 6-15 sec window - yay ReportLab Dev's) - but
since we can't change that (without some major overhauling) we need to look
at code optimization.

I've built everything in memory, trying to use the reportlab-accelerator,
and several other tricks to improve speed.  Early on, I was able to reduce
file sizes from about 300k down to ~120k just by making efficient use of
existing reportlab strategies for dealing with lines (aka Line versus
PolyLine, etc - this also sped-up the report generation considerably).  I'm
looking for advice in the way of "well since you are doing so many lines,
have you thought about ___________", or "you are using OE?  Well, you
should really download and install ________".  But, in the end, if it boils
down to our hardware architecture, then... darn it! =P

We are using an ARMv7, beyond that most of the hardware is unknown to me
(for lots of different reasons).

Got any advice in the form of what I'm looking for?  =P

-J



----


J. R. Carroll
Cell:  (650) 776-6613
www.jrcresearch.net
www.ontvp.com
Email: jrcarroll at jrcresearch.net
          jrcarroll at hurtzlab.com
          jrc.csus at gmail.com
<https://www.facebook.com/J.R.Car>
<https://twitter.com/jNammer><http://www.linkedin.com/in/jrcarroll>



On Wed, Jan 2, 2013 at 3:40 PM, Tim Roberts <timr at probo.com> wrote:


> J. R. Carroll wrote:

> > Throwing this out there, hoping to catch an idea or two:

> >

> > The Setting:  our company has an open embedded device that we have

> > installed reportlab to.  On our local machines, it takes the report

> > less than a second to generate the PDF and save to disk - very quick -

> > very nice (thank you!).  On the target device it's taking anywhere

> > from 6.5sec to 15.sec - not the best case scenario I can think of.

>

> Most embedded devices have woefully underpowered CPUs.  Remember the

> basic purpose of ReportLab is to create Portscript strings from your

> method calls, and accumulate them into one big document.  That means

> vast amounts of string processing, which means lots of memory copies.

> What processor are you using?

>

> --

> Tim Roberts, timr at probo.com

> Providenza & Boekelheide, Inc.

>

> _______________________________________________

> reportlab-users mailing list

> reportlab-users at lists2.reportlab.com

> http://two.pairlist.net/mailman/listinfo/reportlab-users

>

>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://two.pairlist.net/pipermail/reportlab-users/attachments/20130102/b6b7bf74/attachment.html>