Mailing-List: contact user-help@poi.apache.org; run by ezmlm
Precedence: bulk
Reply-To: "POI Users List" <user@poi.apache.org>
Received-SPF: pass (herse.apache.org: domain of rcote@ebi.ac.uk designates
 193.62.196.100 as permitted sender)
Message-ID: <45358.62.56.118.77.1182289078.squirrel@webmail.ebi.ac.uk>
In-Reply-To: <5a59da2e7f47b54b9ce12a2c8bcb7ad7@mac.com>
References: <46781127.6060101@ebi.ac.uk>
    <5a59da2e7f47b54b9ce12a2c8bcb7ad7@mac.com>
Date: Tue, 19 Jun 2007 22:37:58 +0100 (BST)
Subject: Re: HSSFRichTextString.applyFont causing MS Excel to run out of
     memory
From: rcote@ebi.ac.uk
To: "POI Users List" <user@poi.apache.org>
User-Agent: SquirrelMail/1.4.6-1
MIME-Version: 1.0
Content-Type: text/plain;charset=iso-8859-1
Content-Transfer-Encoding: 8bit
Importance: Normal


On Tue, June 19, 2007 8:41 pm, Henry Martin wrote:
>> The problem comes with this last bit of code, when I call
>> hssfTextString.applyFont(). I tried a small test set of data (5 rows,
>> each cell having two or three data items to be processed). It worked
>> perfectly. I then tried it with a production test set (3500 rows, each
>> cell having 4-5 data items). That's when I found that MS Excel could
>> no longer open the document
>
> Can you partition the production test set to find out if it is a
> specific row causing the problem ( either by controlling the source of
> data or by counting rows in your code  )?  A particularly simple test
> might be to use only the first 5 rows from the production test set.

I can, but I have the feeling that's not the issue. I'm thinking object
creation is the problem. I've tried googling the problem and from what I
could find that was reasonably close, it seems that every time you create
an HSSFRichTextString and modify it in some way by applying styles and
fonts, it seems that a new object is created in a shared string area. This
means that if I modify a given cell content a few times with various
substrings, it's creating a whole lot of orphaned SSTs that, to the best
of my knowledge, are not cleaned up. This is purely a wag on my part, but
if that's the problem, I don't know how to get around it.

Anyway, I can try it and post the original xls file (without colors), the
production file (broken, with colors) and a version that has been saved by
OpenOffice on my work website tomorrow if anybody wants to have a better
look.

I appreciate any pointer or comment :)

Cheers,
Rc
-- 
Richard Cote
Software Engineer - PRIDE Project Team (Sequence Database Group)

European Bioinformatics Institute
Wellcome Trust Genome Campus            RCote@ebi.ac.uk
Hinxton
Cambridge CB10 1SD                      Phone: (+44) 1223 492610
United Kingdom                          Fax  : (+44) 1223 494468


---------------------------------------------------------------------
To unsubscribe, e-mail: user-unsubscribe@poi.apache.org
For additional commands, e-mail: user-help@poi.apache.org