Eric Barroca a écrit :
> Hi,
> 
> On 26 avr. 07, at 10:46, Olivier Grisel wrote:
>> We will probably soon need to store checksums of file attachments as
>> part of a customer project. The goal is to be able to quickly find
>> dupes when importing a bunch of files from a file-system folder to a
>> nuxeo workspace.
>> I wondered if it would be relevant to add sha (or md5 ?) checksums by
>> default in nuxeo, either as a Blob new feature or in a dedicated field
>> of the file schema computed by a core event listener.
> Yes it is, as long as the computation of the checksum is done
> asynchronously after an upload.

Asynchronous checksum computation and (transactional) integrity feel a bit
antinomic to me but I am aware that this can be a performance problem and should
thus be configurable.

Another problem is for very large files: the checksum should be computed on the
core side to have a sufficiently low level access to the binary stream so as not
to have to load the content in memory nor to stream it over a remote network
layer of any sort. That's why it might be interesting to have it part of the
Blob default interface or maybe as an extension of it.

> It can definitely add many possible use
> case for the platform especially when dealing with large files.
> So go on for NXP without forgetting to add the corresponding JIRA
> ticket. :-)

Yes sure:

  http://jira.nuxeo.org/browse/NXP-897

-- 
Olivier

_______________________________________________
ECM mailing list
[email protected]
http://lists.nuxeo.com/mailman/listinfo/ecm

Reply via email to