You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@lenya.apache.org by Rudolf Korhummel <ru...@korhummel.de> on 2011/02/26 09:53:02 UTC

uscase editors.insertAssets : heap memory problems

Hi all,

We have a publication with about 1000 assets. If we try to insert new  
assets with the usecase editors.insertAsset from the oneform editor  
the heap raises up 1 GB which runs into out of heap memory when we  
call it a second time and lenya stops working.

I think the problem is that the metadata of all assets have to be read  
to build up the listing.

Here a two suggestions to solve this problem:

1) Use a usecase  like editors.insertLink for images and assets. Only  
the metadata of the child documents have to be read.

2) Provide the information for the listing in the sitetree, so no  
metadata file content is needed.

WDYT

Rudolf
  
  

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


Re: uscase editors.insertAssets : heap memory problems

Posted by Andreas Hartmann <an...@apache.org>.
Am 26.02.11 09:53, schrieb Rudolf Korhummel:
> We have a publication with about 1000 assets. If we try to insert new
> assets with the usecase editors.insertAsset from the oneform editor the
> heap raises up 1 GB which runs into out of heap memory when we call it a
> second time and lenya stops working.
>
> I think the problem is that the metadata of all assets have to be read
> to build up the listing.
>
> Here a two suggestions to solve this problem:
>
> 1) Use a usecase like editors.insertLink for images and assets. Only the
> metadata of the child documents have to be read.

+1

Additionally it would be nice to add a search form for title and keyword 
search.

I'm not sure if the tree structure is the way to go here, though. I 
doubt that many people find a tree-based structure straightforward for 
storing assets. Maybe a keyword/tag-based storage makes more sense? That 
raises the question of tag definitions, though, which is a whole new 
topic altogether.

Anyway, the Dojo DataGrid could be a good starting point for the GUI:
http://dojocampus.org/explorer/#Dojox_Grid_Basic
It supports lazy loading of data via asynchronous requests while scrolling.

Combining the grid for presentation with a live search input field could 
be quite user-friendly. We have done this for another project (although 
with Apache Cayenne, not with Lucene).

> 2) Provide the information for the listing in the sitetree, so no
> metadata file content is needed.

-1

The sitetree in the current form (static in-memory structure, serialized 
and restored as a single XML structure) is already a bottleneck. Making 
it even heavier will further reduce the performance and increase memory 
consumption.

-- Andreas

-- 
Andreas Hartmann, CTO
BeCompany GmbH
http://www.becompany.ch
Tel.: +41 (0) 43 818 57 01


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