----------------------------------------
> From
> To: [email protected]
> Date: Sun, 21 Mar 2010 14:26:38 -0500
> Subject: Re: [iText-questions] Perfomance Question - ByteArray vs Files
>
> This is not about general coding philosophy or the merits of PDFs. I am in
> an iText forum specifically because I need to produce a PDF. I always
> evaluate what the customers needs and wants are to produce the proper
> output. If I didn't want a PDF, I wouldn't be here. End of story.
That doesn't mean you need to produce a PDF de novo everytime someone
from google finds your site, we are talking about optimizations given
constraints. PDF is expensive compared to alt ways to preview for example.
>
> In my opinion, one has to balance coding with resources. I've been on
> servers where developers don't do this, resulting in server crashes and poor
> performance. I don't like it so I always try to understand what is going on
> under the covers to figure out where basic intelligent tradeoffs can be
> made. Yes, there are different techniques that one can use to achieve this
> balance and we can have endless discussions about that. But not now.
>
> I am nowhere near ready to get into advanced techniques in Java or looking
> for bottlenecks buried deep in the server, especially since I'm on a closed
> server with very limited access. Not being a Java programmer and being new
> to iText, I'm at a big disadvantage. I was hoping I could treat iText more
> like a black box and, with a basic understanding, use it efficiently.
>
> I was trying to ask a specific technical question targeted at memory
> utilization of PDFReader and PDFStamper. Seemed pretty obvious to me that
> since I was making two passes there was a tradeoff here. Somewhere we went
> off the tracks.
>
> I am very disappointed there isn't technical expertise available on this on
> forum that can give an overview of the process and answer the question.
>
If you don't have the expertise yourself to evaluate the strategies we have
outlined,
how do you even state with confidence
>[...]I've been on
> servers where developers don't do this, resulting in server crashes and poor
> performance.
maybe they are doing the best they can with the constraints available
and the real problem is you just need to buy a bigger server?
Nobody here knows anything about the statistics or parameters of your system
or data. If you want some general ideas to discuss with your own experts,
I hope we could help. Otherwise, you will have to hope someone who has looked
at the relevant code can just give you a simple answer.
I don't even remember your specific question but what exactly would
you do with the answer? The other thing is that people like you often
ask questions that don't address their likely real concerns. Talking
around a little, time permitting, may help solve an underlying
problem or answer a question someone else has while browsing this in the
archives.
Sometimes people do code kluges for the sake of pushing things out the door.
Maybe you have spotted something that is in fact a coding placeholder
but again it would be easier for someone to just look at the source
code or a heap dump than asking such a question. If I still had the code
somewhere I'd be tempted to look but still not sure what you would do
with an answer. Do you just want someone to read the code to you?
Perhaps you could try asking the original question again.
Not that it would help, but I would point out that multiple passes
through a dataset are generally bad due to loss of memory locality.
This invites thrashing- usually just lower level memory caches but
it can make your disk light stick on due to VM thrashing. The strategy
is to try to do block oriented operations in sizes so that you only use things
in lower level caches.
>
_________________________________________________________________
Hotmail: Trusted email with powerful SPAM protection.
http://clk.atdmt.com/GBL/go/210850553/direct/01/
------------------------------------------------------------------------------
Download Intel® Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
_______________________________________________
iText-questions mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/itext-questions
Buy the iText book: http://www.1t3xt.com/docs/book.php
Check the site with examples before you ask questions:
http://www.1t3xt.info/examples/
You can also search the keywords list: http://1t3xt.info/tutorials/keywords/
