On Thu, Dec 19, 2002 at 02:00:48AM -0500, Gianni Johansson wrote:
> Hi Matthew, Bombe:
>
> I checked in an initial version of the the SplitFileRequestServlet
> built around AutoRequester. It is less polished than I would like,
> but I think it's good enough so you can start working on cleaning up
> UI.
COOL!
>
> To try it out uncomment the lines:
>
> //if (splitFile != null) {
> // // REDFLAG: Need to reconstruct query values in redirect.
> // resp.sendRedirect("/servlet/SFRequest/" + uri);
> // return;
> //}
>
> in FproxyServlet.doGet and add the config lines in the comments above
> them to your freenet.conf.
Okay...
>
> It does work. It hasn't been extensively tested, especially not with
> large multi-segment SplitFiles.
I'm sure this can be arranged :).
>
> Open Issues:
> - anonymity filtering isn't implemented.
> There are already hooks for it. See SplitFileRequestServlet.filterData().
> I think Matthew can probably put this in relatively quickly.
Ok.
> - It downloads all the data from freenet before sending any of it. i.e. it
> doesn't send the data per segment like the current fproxy code does. It
> probably wouldn't be that hard to put the per segment streaming back in.
You want me to look into this?
> - retrying of the initial request for the SplitFile metadata isn't
> implemented. I don't think this will be an issue, because normally the data
> should be cached, Since SFRS only gets a request after fproxy has
> successfully retrieved the SplitFile metadata and sent it a redirect.
I concur.
> - Integration with fproxy needs to be cleaned up. i.e. the redirect url
> shouldn't be hard coded and query parmeters need to be reconstructed for the
> redirect.
> - the dbr date override query argument isn't supported yet.
Ok. These two are easy enough.
> - it doesn't set caching headers like fproxy yet.
This is pretty easy. Will fix.
> - non FEC SplitFile downloads are not supported. Since non-redundant
> Splitfiles almost never work, this isn't a very high priority for me.
Ouch. I don't think we should _insert_ non-redundant splitfiles, but I
see absolutely no reason to abandon support for _fetching_ them. When I
get around to this I will fix it... I don't think we can ship 0.5.1 with
the new splitfile code without the ability to fetch non-FEC splitfiles.
I also think we ought to make sure that nonredundant splitfiles with
different sized blocks and URIs not necessarily CHKs should be tested.
> - the skip local data store option isn't implemented.
Ok.
>
> Known Bugs:
> Canceling sometimes doesn't work. I am looking into this. It looks
> like AutoRequester is swallowing InterruptedExceptions...
Is this the cause of the bug where it hangs after fetching all the
blocks, waiting for a few threads that never return and eventually
timing out and failing despite having enough data to reconstruct?
>
> I would like to hand off the open issues except for the canceling bug
> and the local data store skipping to Matthew and concentrate on
> getting SplitFile insert support into AutoRequester. Of course I will
> also work on fixing any bugs found in the new SplitFile downloading
> code.
Sounds good.
>
> What do you think?
>
> I don't have that much spare programming time. I'm afraid that if I try to
> do all of it myself, I will end up slowing down the next point
> release.
Yeah, we figured this. Any help you need will eventually be forthcoming.
>
> I'll try to catch up with you on IRC in the next couple of days.
Looking forward to it.
>
> --gj
--
Matthew Toseland
toad at amphibian.dyndns.org
amphibian at users.sourceforge.net
Freenet/Coldstore open source hacker.
Employed full time by Freenet Project Inc. from 11/9/02 to 11/1/03
http://freenetproject.org/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL:
<https://emu.freenetproject.org/pipermail/devl/attachments/20021219/0620273d/attachment.pgp>