Am 12.12.2005 um 21:34 schrieb Benji York:

Andreas Jung wrote:
Log message for revision 40683:
- ported OFS.content_types from Z2 to Z3 (which is the most recent maintained code)

As a result of a recent sprint, ZC has a package (zope.mimetype) that would probably be more useful to people interested in such things. I put it and another package that depends on it (zope.file) into the .org repository as top-level projects today.

Reading with interest the code of zope.file I saw, that although the iterator is used for the delivery of the data, the data is first read into memory completly. Also the storage in the zodb is done as one big chunk. Actually it is copied twice before it is wrapped in the iterator.

        data ="rb").read()
        self.headers += ("Content-Length", str(context.size)),
        self.body = bodyIterator(cStringIO.StringIO(data))

Is this done on purpose? Or is this implementation meant as a basic one, which will be replaced by something like a filesystem storage or the new zodb-blob support in the upcoming next version?

With regards,


Zope3-dev mailing list

Reply via email to