shivshan wrote:
Hi
There was another discussion on this topic, where an effort has been made to
store the files directly on the filesystem. Not sure how this would reduce
memory consumption. Please read Sergiu's comments carefully though, before
you try this out.
Link
http://www.nabble.com/AttachmentStore---my-filesystem-based-implementation-…
here
Of course, in the meantime, we await the WebDAV implementation.
Shiva
I'm not sure the webdav implementation will solve the memory problem. If I understand
correctly how
it will work, then it will just be an interface over the same core, meaning that all the
performance
issues will still be there.
vmassol wrote:
>
> On Jun 12, 2008, at 2:45 PM, Artem Melentyev wrote:
>
>> Hi, Vincent.
>>
>> Vincent Massol wrote:
>>> Why is it using some much memory? Isn't that a bug? (I don't
>>> understand why we would need to load more than 1 attachment version
>>> at
>>> a time in memory).
>> Look at
>>
http://xwiki.markmail.org/search/?q=attachments+storage+profiling#query
>> :attachments%20storage%20profiling+page:1+mid:5avjhiu2nkg4xtzt+state:results
>>> Worse, if this is not the first version of the attachment, then the
>>> complete attachment history is loaded in memory, so add another
>>> 24*versionsize*versions of memory needed during upload.
>> This is a feature of JRCS :)
> I see. So I guess the only solution would be drop JRCS and do our own
> stuff... (or find another library that does it fast).
>
> Thanks
> -Vincent
>
--
Sergiu Dumitriu
http://purl.org/net/sergiu/