Tony Nelson wrote:
At 12:09 PM -0800 11/21/07, Ed Swierk wrote:
On 11/21/07, Tony Nelson <tonynelson@xxxxxxxxxxxxxxxxx> wrote:
I'm new to mod_python, but it is documented to be able to do /anything/
that could be a response from Apache. If the handler were to fetch the
data itself it could write a local file and also return the data to the
client, by calling urlopen() and then read(), write(), and req.write()
repeatedly. I don't know if the reads and writes will overlap.
I just implemented this in InstantMirror.
Way to go!
The complication I feared
was handling byte range requests properly. But I don't handle them at
all (i.e. they get treated as full file requests), and yum/urllib
doesn't complain.
Hmm. I expect that will affect resuming downloads. Hopefully somebody who
knows more about serving httpd will speak up.
What currently happens if one instance of InstantMirror.py is
downloading file FOO, and during that download a second client requests
the same file?
Warren Togami
wtogami@xxxxxxxxxx
--
fedora-devel-list mailing list
fedora-devel-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/fedora-devel-list
