Send Devl mailing list submissions to
devl at freenetproject.org
To subscribe or unsubscribe via the World Wide Web, visit
http://www.uprizer.com/mailman/listinfo/devl
or, via email, send a message with subject or body 'help' to
devl-request at freenetproject.org
You can reach the person managing the list at
devl-admin at freenetproject.org
When replying, please edit your Subject line so it is more specific
than "Re: Contents of Devl digest..."
Today's Topics:
1. RE: Proposal: algorithm for forgetting documents in datastore (Neil
Barsema)
2. Re: GMP/GCJ update. (Mark J. Roberts)
3. 0.3.7, wherefore art thou 0.3.7? (Tavin Cole)
4. Re: GMP/GCJ update. (Mark J. Roberts)
5. Re: GMP/GCJ update. (Mark J. Roberts)
6. Re: Aardvark (Oskar Sandberg)
7. Re: Is it just me, or is Freenet sick at the moment? (Adam Langley)
8. Re: Is it just me, or is Freenet sick at the moment? (Oskar Sandberg)
9. Re: Proposal: algorithm for forgetting documents in datastore (Tavin Cole)
10. Re: Announcement Protocol (Tavin Cole)
11. Re: Aardvark (Ian Clarke)
12. Re: Is it just me, or is Freenet sick at the moment? (Adam Langley)
13. RE: Proposal: algorithm for forgetting documents in datastore (Neil
Barsema)
--__--__--
Message: 1
From: "Neil Barsema" <n...@barsema.org>
To: <devl at freenetproject.org>
Subject: RE: [freenet-devl] Proposal: algorithm for forgetting documents in
datastore
Date: Tue, 30 Jan 2001 23:02:06 +0100
Reply-To: devl at freenetproject.org
Tavin Cole wrote:
>But files that are *not* close to the node's keyspace will tend to
>drop out of the cache (if they ever make it in there) because requests
>for those keys won't be routed to that node. So what could be gained
>from applying extra rewards to the close files?
hm,there is no 'close', only 'closer'...
Anyway what I am hoping to achieve is that if a file is popuplar enough to
get moved to a near perfect location
it has a pretty good chance of persisting.
We will never be able to, or should want to guaranty that a file will
persist on the network forever, however if the only way a file can survive
is if it gets hits, Freenet will become a very fleeting network indeed.
Limited recources is not an excuse, because given enough nodes the total
storage space should be phenomenal. So we need a way for files to reach a
place where they can be relativly safe from this hit requirement. The best
place for this would be on the node that is seen by the network to be the
epicentre for that file as this is where the file will move to naturally
anyway.
So to answer your question what would be gained is that the close(r) files
lose some of their hit requirement for persitecne in the network. And the
closer they are to the percieved keyspace centre of the node the smaller the
hit requirement for persistence.
Neil
--__--__--
Message: 2
Date: Tue, 30 Jan 2001 16:57:42 -0500 (EST)
From: "Mark J. Roberts" <m...@statesmean.com>
To: <devl at freenetproject.org>
Subject: Re: [freenet-devl] GMP/GCJ update.
Reply-To: devl at freenetproject.org
On Tue, 30 Jan 2001, Ian Clarke wrote:
> > BTW, the GCJ in CVS now correctly compiles Fred without any patching. The
> > BigInteger support is still broken, however.
>
> Beautiful, I can't wait for our first fully-working Fred binary!
Anyone know how to emulate Java's unsigned right bitshift operator (>>>
and >>>=) in C++? Some info:
Java has also added the unsigned right shift >>>, which uses zero
extension: regardless of the sign, zeroes are inserted at the higher-order
bits. This operator does not exist in C or C++.
--
Mark Roberts
mjr at statesmean.com
--__--__--
Message: 3
Date: Wed, 31 Jan 2001 01:29:00 -0500
From: Tavin Cole <ta...@mailandnews.com>
To: devl at freenetproject.org
Subject: [freenet-devl] 0.3.7, wherefore art thou 0.3.7?
Reply-To: devl at freenetproject.org
Just wondering what was up with 0.3.7 (no pressure :-)
--
// Tavin Cole
--__--__--
Message: 4
Date: Wed, 31 Jan 2001 03:54:07 -0500 (EST)
From: "Mark J. Roberts" <m...@statesmean.com>
To: <devl at freenetproject.org>
Subject: Re: [freenet-devl] GMP/GCJ update.
Reply-To: devl at freenetproject.org
On Tue, 30 Jan 2001, Mark J. Roberts wrote:
> On Tue, 30 Jan 2001, Ian Clarke wrote:
>
> > > BTW, the GCJ in CVS now correctly compiles Fred without any patching. The
> > > BigInteger support is still broken, however.
> >
> > Beautiful, I can't wait for our first fully-working Fred binary!
>
> Anyone know how to emulate Java's unsigned right bitshift operator (>>>
> and >>>=) in C++? Some info:
>
> Java has also added the unsigned right shift >>>, which uses zero
> extension: regardless of the sign, zeroes are inserted at the higher-order
> bits. This operator does not exist in C or C++.
Never mind. It's just a matter of typecasting.
Otherwise progress is good. There's a bug in one of the higher level
methods, BigInteger:divide, so the results for any division-based
operation are wrong. Otherwise, it's coming along: the GMP support works
fully, but my translation of the Java code into C++ has a bug or four.
--
Mark Roberts
mjr at statesmean.com
--__--__--
Message: 5
Date: Wed, 31 Jan 2001 04:28:45 -0500 (EST)
From: "Mark J. Roberts" <m...@statesmean.com>
To: <devl at freenetproject.org>
Subject: Re: [freenet-devl] GMP/GCJ update.
Reply-To: devl at freenetproject.org
On Tue, 30 Jan 2001, Adam Langley wrote:
> On Tue, Jan 30, 2001 at 12:36:58PM -0600, Steven Hazel wrote:
> > > Okay. This seems rather easy, and I could probably do the bulk of
> > > the work, but I need someone fluent in C to check it
> > > out. Volunteers?
> >
> > I can help you with the C.
>
> I'll certainly help if I can.
It looks like the speed of the translated Java->C++ code is roughly
comparable to IBM's implementation, except for very easy operations like
add where the fractional millisecond overhead of calling a couple extra
functions slows it to 150% IBM's speed.
With GMP enabled, performance in one case was 10 times IBM's. ;-)
Still a bug in my translation though. Working on it.
--
Mark Roberts
mjr at statesmean.com
--__--__--
Message: 6
Date: Wed, 31 Jan 2001 11:31:09 +0100
From: Oskar Sandberg <md98-...@nada.kth.se>
To: devl at freenetproject.org
Subject: Re: [freenet-devl] Aardvark
Reply-To: devl at freenetproject.org
Speaking of which, I very nearly had a screaming fit when I saw that the
link from the fproxy page was to a KSK. How fucking stupid can we get?
On Tue, Jan 30, 2001 at 11:24:44AM -0800, Ian Clarke wrote:
> I think that given that Aardvark has been down for the last few days, we
> should remove it from the FProxy gateway page for 0.3.7.
>
> Any suggestions for an alternative?
>
> Ian.
--
'DeCSS would be fine. Where is it?'
'Here,' Montag touched his head.
'Ah,' Granger smiled and nodded.
Oskar Sandberg
md98-osa at nada.kth.se
--__--__--
Message: 7
Date: Wed, 31 Jan 2001 03:04:49 -0800
From: Adam Langley <a...@linuxpower.org>
To: devl at freenetproject.org
Subject: Re: [freenet-devl] Is it just me, or is Freenet sick at the moment?
Reply-To: devl at freenetproject.org
On Wed, Jan 31, 2001 at 01:11:20AM +0100, Oskar Sandberg wrote:
> More importantly, the fact that you are describing the behavior of other
> nodes in ways that none of us has been able to reproduce makes it next to
> impossible to understand what it going on.
I've had it happening with a local node, where I can see what it thinks
it's sending (via fred's logs). It's just that the kernel never knows anything
about it. I've got the logs of it at home somewhere.
AGL
--__--__--
Message: 8
Date: Wed, 31 Jan 2001 12:15:37 +0100
From: Oskar Sandberg <md98-...@nada.kth.se>
To: devl at freenetproject.org
Subject: Re: [freenet-devl] Is it just me, or is Freenet sick at the moment?
Reply-To: devl at freenetproject.org
On Wed, Jan 31, 2001 at 03:04:49AM -0800, Adam Langley wrote:
> On Wed, Jan 31, 2001 at 01:11:20AM +0100, Oskar Sandberg wrote:
> > More importantly, the fact that you are describing the behavior of other
> > nodes in ways that none of us has been able to reproduce makes it next to
> > impossible to understand what it going on.
>
> I've had it happening with a local node, where I can see what it thinks
> it's sending (via fred's logs). It's just that the kernel never knows anything
> about it. I've got the logs of it at home somewhere.
If Fred thinks it sent it, then it wrote the message to the socket. If you
still are not seeing it, then it is not a bug in Fred but either in java
or somewhere else.
>
> AGL
>
> _______________________________________________
> Devl mailing list
> Devl at freenetproject.org
> http://www.uprizer.com/mailman/listinfo/devl
--
'DeCSS would be fine. Where is it?'
'Here,' Montag touched his head.
'Ah,' Granger smiled and nodded.
Oskar Sandberg
md98-osa at nada.kth.se
--__--__--
Message: 9
Date: Wed, 31 Jan 2001 11:52:12 -0500
From: Tavin Cole <ta...@mailandnews.com>
To: devl at freenetproject.org
Subject: Re: [freenet-devl] Proposal: algorithm for forgetting documents in
datastore
Reply-To: devl at freenetproject.org
On Tue, Jan 30, 2001 at 11:02:06PM +0100, Neil Barsema wrote:
> Tavin Cole wrote:
> >But files that are *not* close to the node's keyspace will tend to
> >drop out of the cache (if they ever make it in there) because requests
> >for those keys won't be routed to that node. So what could be gained
> >from applying extra rewards to the close files?
>
> hm,there is no 'close', only 'closer'...
>
> Anyway what I am hoping to achieve is that if a file is popuplar enough to
> get moved to a near perfect location
> it has a pretty good chance of persisting.
> We will never be able to, or should want to guaranty that a file will
> persist on the network forever, however if the only way a file can survive
> is if it gets hits, Freenet will become a very fleeting network indeed.
> Limited recources is not an excuse, because given enough nodes the total
> storage space should be phenomenal. So we need a way for files to reach a
> place where they can be relativly safe from this hit requirement. The best
> place for this would be on the node that is seen by the network to be the
> epicentre for that file as this is where the file will move to naturally
> anyway.
> So to answer your question what would be gained is that the close(r) files
> lose some of their hit requirement for persitecne in the network. And the
> closer they are to the percieved keyspace centre of the node the smaller the
> hit requirement for persistence.
The popularity of any file is intrinsically weighted by its closeness to
the node's keyspace focus, b/c of the way routing works. So if we have
a group of files that are all close to the keyspace focus, and we have
to delete one of them, what can we do but delete the one with the
fewest hits?
--
// Tavin Cole
--__--__--
Message: 10
Date: Wed, 31 Jan 2001 11:56:04 -0500
From: Tavin Cole <ta...@mailandnews.com>
To: devl at freenetproject.org
Subject: Re: [freenet-devl] Announcement Protocol
Reply-To: devl at freenetproject.org
On Wed, Jan 31, 2001 at 01:06:56AM +0100, Oskar Sandberg wrote:
> My parser can read FNP (and html without modification it turns out) from
> any stream, can't yours?
^^^^^^^^^^^^^^^^^^^^^^^^^
oh really? great, we can put the content filter at the protocol level!
--
// Tavin Cole
--__--__--
Message: 11
Date: Wed, 31 Jan 2001 09:28:07 -0800
From: Ian Clarke <i...@octayne.com>
To: devl at freenetproject.org
Subject: Re: [freenet-devl] Aardvark
Reply-To: devl at freenetproject.org
--Sr1nOIr3CvdE5hEN
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable
On Wed, Jan 31, 2001 at 11:31:09AM +0100, Oskar Sandberg wrote:
>=20
> Speaking of which, I very nearly had a screaming fit when I saw that the
> link from the fproxy page was to a KSK. How fucking stupid can we get?
I put that there as an exercise for the reader.
Ian.
--Sr1nOIr3CvdE5hEN
Content-Type: application/pgp-signature
Content-Disposition: inline
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.4 (GNU/Linux)
Comment: For info see http://www.gnupg.org
iD8DBQE6eEsnQtgxRWSmsqwRAvqYAJ9nDuA3Jy8bAiq+mvdeIq9tWGAU/wCfWGGb
8ioKYBEyICFxK/tKm5aaPlo=
=5yKv
-----END PGP SIGNATURE-----
--Sr1nOIr3CvdE5hEN--
--__--__--
Message: 12
Date: Wed, 31 Jan 2001 18:45:44 +0000
From: Adam Langley <a...@linuxpower.org>
To: devl at freenetproject.org
Subject: Re: [freenet-devl] Is it just me, or is Freenet sick at the moment?
Reply-To: devl at freenetproject.org
--yudcn1FV7Hsu/q59
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable
On Wed, Jan 31, 2001 at 12:15:37PM +0100, Oskar Sandberg wrote:
> > I've had it happening with a local node, where I can see what it thinks
> > it's sending (via fred's logs). It's just that the kernel never knows a=
nything
> > about it. I've got the logs of it at home somewhere.
>=20
> If Fred thinks it sent it, then it wrote the message to the socket. If you
> still are not seeing it, then it is not a bug in Fred but either in java
> or somewhere else.
Quite possibly.
AGL
--=20
The difference between genius and stupidity is that genius has its limits.
--yudcn1FV7Hsu/q59
Content-Type: application/pgp-signature
Content-Disposition: inline
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.4 (GNU/Linux)
Comment: For info see http://www.gnupg.org
iEYEARECAAYFAjp4XVgACgkQzaVS3yy2PWBxKQCgqmpoAljNx6JnCWvfQznqpgks
fwcAn31t/M/IAtooCeXKF52/Ue8X//SA
=ACda
-----END PGP SIGNATURE-----
--yudcn1FV7Hsu/q59--
--__--__--
Message: 13
From: "Neil Barsema" <bars...@pno.org>
To: <devl at freenetproject.org>
Subject: RE: [freenet-devl] Proposal: algorithm for forgetting documents in
datastore
Date: Wed, 31 Jan 2001 20:05:55 +0100
Reply-To: devl at freenetproject.org
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
> Tavin Cole wrote:
>The popularity of any file is intrinsically weighted by its
>closeness to the node's keyspace focus, b/c of the way routing
>works. So if we have a group of files that are all close to the
>keyspace focus, and we have to delete one of them, what can we do
>but delete the one with the
>fewest hits?
This is not what is happening now, the file with the longest wait
since its last hit is deleted.
I would rather delete the one that got few hits but is also likely to
be cashed elsewhere.
Remember in my scheme getting hits is still a much more powerfull way
to avoid deletion for file than being instrumental in routing
decisions.
The latter is just an indication that the file is on the 'right'
node. not just an echo from some request still with a high htl.
Neil
-----BEGIN PGP SIGNATURE-----
Version: PGPfreeware 6.5.8 for non-commercial use <http://www.pgp.com>
iQA/AwUBOnhUA5FMYYlPEhpIEQIy0gCgwTHxsEf/+UIdJl9jPTOvirZ3aVoAoPoX
EJ3jrRlfydiRNrgBLuIf7A9D
=Syne
-----END PGP SIGNATURE-----
--__--__--
_______________________________________________
Devl mailing list
Devl at freenetproject.org
http://www.uprizer.com/mailman/listinfo/devl
End of Devl Digest
