[lugdunummaster]

Quote

Precisely! I mean, is it so hard to send a message listing the filter(s) matching the query?

Nope. In fact this is implemented since ages.

Try to publish a file with ketamine in its filename on DsNo1

Then again, servers owners are... server owners. Not guys that may satisfy all your past, current and future dreams.

filtering is done at publish time only.

So no message is sent at retrieval time. Would be quite useless and very expensive to include in a search answer a message like :

NNNN files were filtered because of filter[0]=(#FORMAT dll)|(#FORMAT sys)
PPPPPP files were filtered because of filter[1]=(#SIZE<5)
...
XXXXX files were filtered because of filter[78]=!(.)
CCCCCC files where rejected because of their hash (whatever name was published)

Because even if you searched for "wikipedia", results may miss "The king of wikipedia" that you really wanted to fetch. No ?

Server owners have the tools, they are free to use them the way they want.

[tHeWiZaRdOfDoS]

Do you store filtered publishes anyways?
If you do, then then it's clear that you can't answer on such searchrequests, neither can you provide a "filtered" answer...
If not, it might save some RAM plus reduce the costs for the lookups on search requests if you remove filtered content from the server-DB and/or don't add it at all...

[EvolutionCrazy]

tHeWiZaRdOfDoS, on Aug 10 2006, 01:34 AM, said:

muleteer, on Aug 9 2006, 09:53 AM, said:

Could you please provide some support for a server credits meter or blacklisting status indicator? Send the credits info to the client, eg, during the periodic server connection refresh.

That wouldn't help much but only cause unnecessary overhead...
However, Lugdunumaster, some time ago I sent you a PN with a similar context though I never received an answer, so I abuse this thread now and ask again:
Could you please provide a "price list" for every user action? e.g. "Publishing files costs X (base) + n(number of files)*Y (cost per file)" - TCP/UDP
There once was a list on "SilentBob" but it lacked great parts plus it's too old, now...
That way modders could write a proper "respect server credits" feature to e.g. limit searchresults or whatever and to prevent getting banned by a server.
I'm currently estimating most values but I still get banned sometimes... especially after completing several small files pretty fast as the filelist is then updated and sent to the server again and again.

GreetZ,
WiZ

neomule mod already implement that resource estimator, check it out

this is maybe the page you were referring to:
http://www.der-still...k...ist&lang=en

[tHeWiZaRdOfDoS]

NeoMule copied an ancient code from TheBlackHand Mod and modified it to work on 0.47a - I know that and I knew it for a long time now, it's neither reliable nor acurrate nor up2date - that's why I were asking Lug.
And yes that's the mentioned page...

[PacoBell]

lugdunummaster, on Aug 8 2006, 09:45 PM, said:

eserver has the ability to filter by hashes since ages. And very fast. list of 100.000.000 hashes is OK.

lugdunummaster, on Aug 9 2006, 11:23 PM, said:

Quote

Precisely! I mean, is it so hard to send a message listing the filter(s) matching the query?

Nope. In fact this is implemented since ages.

By the way, where is this all documented? I can't find any mention of this on the usual page.

[EDIT]
Well, I see warnfakes listed, if that's what you were referring to.
[/EDIT]

This post has been edited by PacoBell: 10 August 2006 - 10:09 AM

[lugdunummaster]

Oh dear

The day emule respects the softlimit of the server, instead of publishing the whole files shared by the user, you can ask me to add features.

Right now emule ignores (and always ignored) the message sent by the eserver (softLimit reached) and happily continue to send chunks of 200 files (and consumes credits for nothing)

[tHeWiZaRdOfDoS]

Lugdunummaster I think if you would provide a better overview about the Server-protocol then a lot of problems could be solved... I dunno whether you talk to the official devs about the software, the used opcodes and such but I've never seen a proper overview for other devs

[PacoBell]

lugdunum said:

The server now sends a message to the user when the soft limit is reached "WARNING : You have too many shares for this server".
This message can be modified by nbuser

Did you mean this the message or is there a more standard opcode that's sent out?

This post has been edited by PacoBell: 10 August 2006 - 01:50 PM

[CiccioBastardo]

lugdunummaster, on Aug 10 2006, 08:23 AM, said:

Quote

Precisely! I mean, is it so hard to send a message listing the filter(s) matching the query?

Nope. In fact this is implemented since ages.

Try to publish a file with ketamine in its filename on DsNo1

Then again, servers owners are... server owners. Not guys that may satisfy all your past, current and future dreams.

filtering is done at publish time only.

So no message is sent at retrieval time. Would be quite useless and very expensive to include in a search answer a message like :

NNNN files were filtered because of filter[0]=(#FORMAT dll)|(#FORMAT sys)
PPPPPP files were filtered because of filter[1]=(#SIZE<5)
...
XXXXX files were filtered because of filter[78]=!(.)
CCCCCC files where rejected because of their hash (whatever name was published)

Because even if you searched for "wikipedia", results may miss "The king of wikipedia" that you really wanted to fetch. No ?

Server owners have the tools, they are free to use them the way they want.

Sorry, I missing something.
How can a single string be more expensive of what should be a long list of sources + metadata?
Moreover I can't see it useless. If I know server X filters my requests for "snorky" I'll never try to make a query for those words on it (global searching is different, but with a feedback from a server it is possible to exclude the filtering server from the search, I belive (less traffic, less credits used)).
And no, the feedback on the search should not be made on filtered content (as, if you have already blocked it at publish time, you should not know it ever existed, unless, as Wizard suggested, you have a separate list for them). You send a warning based on query words.: king AND wiki -> send a warning wiki NOT king -> do not send a warning

What could be wrong in this?

[pepe0008]

Hmm, kind of off-topic and newbie question... but...

Why is not open-sourced the lugdunum eserver software? (apart from because his author doesn't want it to be, that is).

Well, I guess the question should be: why doesn't lugdunum want to release his edonkey server as open-source?

Thanks for reading.

[strip]

Server program has never been opensourced. The now defunct metamachines company gave the source to lugdunummaster. I do not know under what conditions.

It would be interesting to know if these conditions will suffer any legal change. Who is the owner now?

This post has been edited by strip: 24 September 2006 - 09:53 AM

[pepe0008]

strip, on Sep 24 2006, 11:49 AM, said:

Server program has never been opensourced. The now defunct metamachines company gave the source to lugdunummaster.

Yes, that "seems" to be the case, also according to this web interview.

However, says lugdunum in this very thread:

Quote

Well.. it took me one month to write first eserver.

Therefore I am under the impression that Lugdunum had access to the original MetaMachine edonkey server sources, but that he wrote his own version of an edonkey server from scratch with the knowledge thus acquired.

Did MetaMachine impose any legal restriction(s) on Lugdunum for his new edonkey server version? If so, what can we expect after MetaMachine recently trying to shut down the edonkey network?

Is open-sourcing the Lugdunum edonkey server possible? Does Lugdunum want to do it?

What is the rationale behind whatever decision is taken?

Inquiring minds want to know...

[lugdunummaster]

Quote

Therefore I am under the impression that Lugdunum had access to the original MetaMachine edonkey server sources, but that he wrote his own version of an edonkey server from scratch with the knowledge thus acquired.

Did MetaMachine impose any legal restriction(s) on Lugdunum for his new edonkey server version? If so, what can we expect after MetaMachine recently trying to shut down the edonkey network?

Is open-sourcing the Lugdunum edonkey server possible? Does Lugdunum want to do it?

What is the rationale behind whatever decision is taken?

Inquiring minds want to know...

Original dserver was written by Metamachine, but could not handle more than 4000 users, because of poor design (and sources being the same for window and unix)

I did a complete reverse engineering of this program and network sniffing (as did emule for the client).

I did many patches in assembly in the binary image of dserver. Servers were then able to reach 100.000 users.

Then, assuming Metamachine was not interesting into maintaining themselves dserver (they were working on Overnet, wich was not using dservers anymore), I asked them if I could have a copy of sources so that I could go one step ahead.

Metamachine agreed and gave me a copy of their sources. As they were in C++ and with too many levels of abstraction (nice for the programmers, but performance killers), I decided to just delete them all and not spend any time on them. Sometime, it's better to throw the whole thing in the bin and not pollute your head with wrong design.

Later, when it became obvious that edonkey/emule network still needed big servers because Kademlia clients were not yet ready, I wrote a 100% lugdunum implementation, in C language and with the right design. This was three years ago (Sep 2003).
This C version let me add new protocol extension to reduce overall bandwith (zip compression, udp requests aggregation), and to reduce RAM usage with special allocators.

I own eserver 100%, and it is not open source because many bad guys would use the source to build fake servers. Fake clients already exist and harm the network.

[Firnus]

lugdunummaster, on Sep 25 2006, 06:54 AM, said:

because many bad guys would use the source to build fake servers

Fake server still exist without your source.

And there are fake edonkey client based, even if it's closed source. I don't think close source can protect efficiently.

But it's your choice

(Thanks for all)

This post has been edited by Firnus: 25 September 2006 - 08:14 AM

[lugdunummaster]

Quote

Fake server still exist without your source.

Sure, but they are obvious fakes and only stupid users may still be fooled by them.

This kind of argument is quite wrong indeed.

If my sources were available, even a bad programmer could easily add stuff and claim it's a lugdunum server, and harm the network.

Without sources, you have to be a really good programmer to do a clone. So far some programmers tried and failed.

Consider if you had to do an emule clone, without access to emule sources.
How many days do you think you'll have to work before having a working thing ?

[Firnus]

lugdunummaster, on Sep 25 2006, 09:04 AM, said:

Consider if you had to do an emule clone, without access to emule sources.
How many days do you think you'll have to work before having a working thing ?

Even, with the source, i never tried ....

But i think RIAA/MPAA can pay really good developpeur an ingenner , no ?

I think actually the fake server is good for what they want, and don't need more developpement into (surely Log développement , not in Lugdunum server)
I think the obvious fake of 'fake' server is to discourage people to use eMule, not to had a good server.

I don't see who can try to make a good server and can be bad for eMule community, but .... maybe ... ??

[EvolutionCrazy]

lugdunummaster, on Sep 25 2006, 08:54 AM, said:

I own eserver 100%, and it is not open source because many bad guys would use the source to build fake servers. Fake clients already exist and harm the network.

that kind of choice is welcomed in most of the cases

@Firnus what do you think will happen if starting from now someone invent a new client that work in a new network, everything cyphered and hardly renversable by just sniffing the network traffic, and register it as a patent deny ANY kind of usage/modification and so on?
What will happen when a RIAA member will show up in court that they have a modified client used to track the users?
At least the would be sued for patent/trademark/license violation and nobody would be able to certify that the logs produced by their software is correct and not wrongly builded

[Firnus]

EvolutionCrazy, on Sep 26 2006, 11:11 AM, said:

@Firnus what do you think will happen if starting from now someone invent a new client that work in a new network, everything cyphered and hardly renversable by just sniffing the network traffic, and register it as a patent deny ANY kind of usage/modification and so on?

You don't think client like this still exist ?
There are a legal problem in france with 2 society who use a program like this ...

Quote

What will happen when a RIAA member will show up in court that they have a modified client used to track the users?
At least the would be sued for patent/trademark/license violation and nobody would be able to certify that the logs produced by their software is correct and not wrongly builded

RIAA had already Fake server, nobody would be able to certify bla bal ...
They don't need to had a good server, just i explain, the only think they want it's a serveur with good Log (send shared , asking source etc ...) AND i'm sure they had sever like this.

[kewlFFT]

lugdunummaster　
It makes a lot of sense, making it public would trigger the apparition of plenty of new servers, some fake, some bugged some good. This could become a big mess.
Respect for your coding of very high quality dedicated to efficiency and speed.

[Firnus]

Heu ....

I nevr say lugdunummaster had to put his job Open-source, he do what he want.

But i say, put his job closed source don't protect from false server.

And lugdunummaster