On Tue, May 25, 2004 at 11:16:00AM +1100, Sergey Volobuev wrote:
The fact that the files are stored within Zope means that you will still have to serve them up via Zope (either download or streaming), and as has been previously mentioned, Zope is not very good at serving up large files.
As i understood from announce, Zope 2.7.1 will contain some improvements in this area.
Sort of. 2.7.1 includes a feature which product developers can use to stream data directly from the filesystem faster than current solutions (and fast enough to saturate a T3, so there's probably no point in going faster :-). But it depends on the product to take advantage of it, by returning a special iterator instead of a string; and - notably - it's unsafe to do database reads inside the iterator. So it's really only good for stuff that is stored or cached on the filesystem. We did write a filesystem cache that uses this feature. That might be what you're thinking of. More info: http://www.slinkp.com/code/zopestuff/blobnotes -- Paul Winkler http://www.slinkp.com