Matthew Toseland wrote:
> - The current FEC decode API is monstrous. We have agreed that the node
> should do all the work, and in 0.7 it should be possible for a very
> simple client to get the node to download a splitfile.
> - We will need to implement metadata parsing, FEC decoding, and so on,
> in the node. We will have to cache a significant amount of data. We
> will have to run the requests. Probably we will have to store the
> whole file. Therefore we will in all likelihood have a download queue
> integrated into the node, which is accessible via FCP.
I'm sorry if I'm complaining a lot.. But I have written two FCP
libraries (one OS/2 specific, and one in Python). Both of them allow me
to do this (read this as pseudo code):
------------
ctx = FcpContext()
ctx.addNode('localhost')
ctx.addNode('foo.baz')
getCtx = ctx.getCtx('CHK at lArgEfILe', 'outfile.dat')
status = getCtx.run()
------------
It will use both nodes (and more if available) to download the file.
I use this a lot, and it has provided really good performance gains.
(And yes, the libraries support FEC encoding/decoding on all the
available nodes, so in my case, two segments can be encoded/decoded at
the same time, on two different computers, for the same file).
The thing I really don't like about FCP programming is the meta data
handling, so I would like to see it go (or better
specifications/documentation of them), but on the other hand, the
current design does allow me to spread fetches among several nodes
(which, I guess, has a extremely minor anonymity role, also).
Please keep the low level/manual interface available as an option for
multinode FCP libraries.
> - Historically we have not included any non-web GUI code in Freenet.
> Partly this was because of the total lack of any free implementation
> of Swing. This is being remedied at some speed by RedHat, but I don't
> know if e.g. Frost works on Kaffe... I do think it must be possible to
> use all the basic functionality of Freenet with Free Software only, if
> this is reasonably easily possible.
Please don't sacrifice resource limiting and performance for the sake
of unfinished Java implementations. I dislike the idea of an application
being suboptimal on my system, because someone else isn't finished with
their clone of Java.
I do agree that free Java implementations should be supported from an
ideological point of view, but face it: In the end, the clones need to
catch up - the applications shouldn't suffer.
If running Freenet with another Java implementation doesn't work, but
is important to a lot of people, then the developers of those Java
implementations will hear about it, and if they care about useability
(and recieve donations), they will do their best to fix it. I hope.
--
Kind Regards,
Jan Danielsson
Te audire no possum. Musa sapientum fixa est in aure.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 187 bytes
Desc: OpenPGP digital signature
URL:
<https://emu.freenetproject.org/pipermail/devl/attachments/20050831/e18249c9/attachment.pgp>
